Western University
Scholarship@Western
Electronic Thesis and Dissertation Repository
January 2016
Applicative Structure in Wolof
Christen Harris
The University of Western Ontario
Supervisor
Dr. Ileana Paul
The University of Western Ontario
Graduate Program in French
A thesis submitted in partial fulfillment of the requirements for the degree in Doctor of Philosophy
© Christen Harris 2015
Follow this and additional works at: http://ir.lib.uwo.ca/etd
Part of the African Languages and Societies Commons
Recommended Citation
Harris, Christen, "Applicative Structure in Wolof " (2015). Electronic Thesis and Dissertation Repository. Paper 3464.
This Dissertation/Thesis is brought to you for free and open access by Scholarship@Western. It has been accepted for inclusion in Electronic Thesis
and Dissertation Repository by an authorized administrator of Scholarship@Western. For more information, please contact [email protected].
APPLICATIVE STRUCTURE IN WOLOF
(Thesis format: Monograph)
by
Christen Harris
Graduate Program in French Studies
A thesis submitted in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
The School of Graduate and Postdoctoral Studies
The University of Western Ontario
London, Ontario, Canada
© Christen Harris 2015
Abstract
This dissertation investigates applicative structures in Wolof, based on new data collected
from native speakers in Saint Louis, Senegal. The dual purpose of this dissertation is to
describe the applicative constructions available in Wolof and to identify their syntactic
structure. Following previous work on applicatives, the description of these applicatives
focuses on their object properties and the c-command configuration of the VP. The
analysis is framed within the Minimalist Program (Chomsky 1993, 1995, 2000). I
propose multiple function heads involved in applicative formation which account for the
properties in Wolof.
Four types of applicatives, benefactive, dative, instrumental, and locative in
Wolof. They are classified into three groups based on their object properties, selectional
restrictions, and c-command configuration. The groups are benefactive applicatives,
dative applicatives, and oblique applicatives (including instrumental and locative). Object
properties will show that benefactive and dative applicatives are symmetrical applicatives
while instrumental and locative applicative, which have been previously identified as
symmetrical (Dunigan 1994), show mixed symmetrical and asymmetrical behaviour. Ccommand tests will show that in benefactive and dative applicatives, the applied object
asymmetrically c-commands the theme but in instrumental and locative applicatives, it is
the theme that asymmetrically c-commands the applied object.
The analysis proposed is based on the Thematic and Raising Applicative
Hypothesis from Georgala (2012). I propose a third applicative head in addition to
thematic and raising Appls, which I call Oblique Appl. Although not a standard
theoretical tool in Minimalism, the notion of Downward Merge from Phillips (2003) and
McGinnis (2005) is incorporated in Oblique Appl to account for instrumental and locative
applicatives which fall outside the explanatory power of the Raising and Thematic
Hypothesis vis-à-vis c-command and verbal adjacency. In the spirit and Marantz (1993)
and Georgala (2012), I argue that all three applicative heads merge in the same position,
!ii
above the lexical VP. I maintain that instrumental and locative applied objects are
uniformly merged as VP-external objects contrary to Marantz who assume they can
merge within the lexical VP.
Keywords: applicative, Wolof, double object construction, verbal valency, instruments,
locatives
!iii
Ndànk ndànk mooy jàpp golo ci ñaay.
“It is slowly that one catches the monkey in the bush.”
Wolof Proverb
!iv
For my father, whose boundless curiosity always inspired me.
Laurence Earl Harris
1955-2015
!v
Acknowledgements
I would first like to thank Ileana Paul, my advisor. Her constant support and
encouragement were instrumental in the achievement of this dissertation. Her
understanding and love of syntax have moulded and motivated me in both the good times
of this process and the bad.
I would also like to thank the professors that taught me and helped me over the
years, David Heap, Jacques Lamarche, Michiya Kawai, Jeff Tennant, and François Poiré.
I would also like that thank my senior thesis advisor, Sarah Gordon, who set me on the
path of research and languages.
I am deeply indebted to Boubacar Boris Diop and his son Moustapha Diop for
taking me in during my fieldwork in Senegal. They introduced me to many Wolof
speakers and linguists as well as providing me with wonderful insight into Wolof. I would
also like to thank Ibrahim Saar and Abdoulaye Dial for their help teaching me Wolof and
understanding its grammar. Finally, I cannot forget the wonderful and generous people
who worked with me as native language consultants including Badara Saar, Dame Diop,
Moussa Ndiaye and Hyacinthe Dione, and Xadija Diene as well as those who wished to
remain anonymous.
I have been lucky to be surrounded by such wonderful and stimulating fellow
graduate students. Thank you Kristen Izaryk for listening to many a syntactic rant and
always taking the time to chat with me about research. I also would like to thank Jorge
Emilio Roses Labarada and Olga Kharytonava who also helped and encouraged me along
the way. I cannot forget Nico Baier with whom I had many delightful and stimulating
discussion about Wolof, applicatives, syntax, and all these Senegal.
Lastly, I cannot forget to thank my wonderful family who have supported me
throughout the entire journey. Darryl Wheeler, who has loved me through the good times
and the bad and was always ready with a hug and a Blizzard. My mother, Ann Marie
!vi
Harmon who was there when I needed countless pep talks and my father, Laurence
Harris, who learned about linguistics just so he could talk to me about my work. He was
always there to listen and to help me think outside the box. Sadly, he was unable to make
it to see this moment, but he will always be in heart.
!vii
Table of Contents
Abstract
ii
Acknowledgements
vi
List of abbreviations
xii
CHAPTER 1
1. Introduction to Wolof
1
1.1. Clause structure
3
1.1.1. Clause Types
4
1.1.2. Nouns and DPs
10
1.1.3. Adjectives
13
1.1.4. Focus and Question Formation
13
1.2. Methodology
14
1.3. Overview of thesis
16
CHAPTER 2
2. Applicatives: Theoretical Background
17
2.1. Applicatives
18
2.1.1. Typology
20
2.1.2. Object properties
24
2.1.2.1. Passivization
25
2.1.2.2. Pronominalization
27
2.1.2.3. Word Order
29
2.1.2.4. Extraction
31
2.1.3. C-command
33
2.1.4. Summary
36
2.2. Syntactic Approaches
36
2.2.1. Issues in applicative research
36
2.2.2. Marantz
38
!viii
2.2.3. Pylkkänen
44
2.2.4. McGinnis
50
2.2.5. Georgala
52
2.2.6. Summary
57
2.3. Existing work on Wolof applicatives
58
2.3.1. Schwartz 1975
58
2.3.2. Dunigan 1994
59
2.3.3. Dione 2013
64
2.4.Theoretical Assumptions
65
2.4.1. Overview of proposed analysis
2.5. Conclusion
66
68
CHAPTER 3
3. Wolof benefactive and dative applicatives
70
3.1. Object Properties
70
3.1.1. Word Order
71
3.1.2. Pronominalization
72
3.1.3. Extraction
76
3.1.4. Object Omission
78
3.1.5. Summary of Object Properties
81
3.2. C-command
82
3.2.1. Tests
82
3.2.1.1. Quantifier Binding
82
3.2.1.2. Reflexivization
90
3.2.1.3. Weak Cross-Over
93
3.2.1.4. Summary of c-command tests
99
3.3. Analysis
99
3.3.1. Benefactives
100
3.3.2. Datives
103
!ix
3.3.3. Variable Word Order
109
3.3.3.1. Object Scrambling
115
3.3.3.2. Different Merges
118
3.3.3.3. Rightward Adjunction
121
3.3.4. A ‘small clause’ problem
122
3.4. Conclusion
126
CHAPTER 4
4. Oblique Applicatives
128
4.1. Oblique objects in Wolof
128
4.1.1. In situ applicatives
130
4.1.2. Hybrid Applicatives
133
4.1.3. Prepositional Instruments and Locatives
134
4.1.4. Fronted Oblique Objects
135
4.2. Object Properties
138
4.2.1. Pronominalization
138
4.2.2. Extraction in oblique applicatives
142
4.3. C-command
147
4.3.1. Quantifier Binding
148
4.3.2. Weak Cross-Over
150
4.3.3. Conclusion
153
4.4. Analysis
153
4.4.1. Lack of Object Scrambling
156
4.4.2. Extraction
158
4.4.3. The in-between nature of instruments and locatives
160
4.4.4. Other applicative proposals
162
4.4.5. Movement out of VP
165
4.5. Conclusion
169
!x
CHAPTER 5
5. Conclusion
170
REFERENCES
174
CURRICULUM VITAE
184
!xi
Abbreviations
ABS - absolutive case
ACC - accusative case
APPL - applicative
ASP - aspect
AT - actor topic
BEN - benefactive
CAUS - causative
CL - classifier
CREL - complementizer of relative clauses
CS - construct state
CT - circumstantial topic
DAT - dative
DEF - definite
DET - determiner
ERG - ergative case
FOC - focus
FV - final vowel
GEN - genitive case
HABIT - habitual
IND - indicative
INST - instrumental
LOC - locative
NOM - nominative case
OFOC - object focus
OM - object marker
OP - object pronoun
PASS - passive
PAT - patient
PERF - perfective
PL - plural
PRES - present
PST - past
REL - relativizer
SA - subject agreement
SFOC - subject focus
SG - singular
SP - subject pronoun
!xii
!1
CHAPTER 1
1. INTRODUCTION TO WOLOF
Wolof is a language of the West Atlantic branch of the Niger Congo family. It is primarily
spoken in Senegal and The Gambia with speakers also found in Mauritania, Mali, and
Guinea-Bissau. As of 2006, Ethnologue estimates approximately four million first
language speakers between these four countries. Estimates of both first and second
language speakers worldwide are around 7 million (Torrence 2012).
Wolof is the main vehicular language in Senegal, and is used, alongside the
official language French, in political venues, domestic business, and the media. Even
with the widespread use, the language remains largely oral. Wolof is spoken by
approximately 80 percent of the population either as a first or second language (Taylor
1995). While French is the official language of government and education, the majority
of the population do not speak it. Thus, Wolof is used for daily communication instead of
French by many Senegalese.
The first major dialect split in Wolof is between Senegalese Wolof and Gambian
Wolof. The two are mutually intelligible but are marked by differences in phonology,
morphology, and syntax. I set Gambian Wolof aside and focus only on Senegalese Wolof.
As for Wolof spoken in Senegal, Ethnologue lists five dialects: Baol, Cayor, Dyolof,
Lebou, and Jander (Lewis et al. 2014). Waalo is another dialect not mentioned by
Ethnologue (Torrence 2012).
!2
Figure 1: Map of Wolof Dialects in Senegal
It is important to note that Lebou Wolof is not the same dialect as Dakar Wolof which
will be described next.
Another dialect which is not mentioned on Ethnologue is Dakar Wolof, also
known as Urban Wolof. Dakar Wolof is an urban dialect spoken primarily in the Dakar
region of Senegal, as the name suggests. It is characterized by borrowed lexical items,
code-switching, and reduction of the nominal class system. French, English, and Arabic
loan words are the main sources of borrowed words and expressions in Dakar Wolof. As
for syntactic characteristics, the most salient variation is the reduction in the noun class
system from fifteen classes to two classes: b- for singular nouns and y- for plural nouns.
The last dialect I would like to mention is the Wolof spoken in Saint Louis and the
immediate surrounding area. Saint Louis is in the north of Senegal on the coast of the
!3
Atlantic Ocean and the southern border of Mauritania. Zribi-Hertz and Diagne
characterize the Wolof of Saint Louis, in sociolinguistic terms, as “upper-class
conservative” (2002:827). They note that speakers in Saint Louis cultivate an awareness
of ‘Good Wolof’. There are fewer borrowed words than Dakar Wolof and all the noun
classes have been retained by speakers. The data analyzed in this thesis come from the
Saint Louis dialect.
Since education in Senegal takes place in French, most speakers are not literate in
Wolof. However, in recent years, there has been a significant push to integrate more
national languages into the school systems leading to Wolof and other national languages
classes being taught in schools. Publication of Wolof newspapers and books as well as
educational material and dictionaries is climbing, but still lags far behind French
publications. A standard orthography for Wolof was adopted in 1972 but spelling in
everyday contexts remains fluid. Most speakers spell phonetically using French
orthography. For example, [u] is spelled with a ‘u’ in standardized Wolof but many
people write [u] with ‘ou’ as in French. I present all data using the standard orthography
as outlined by Diouf in Dictionnaire wolof-francais (2003) and Fal (1999).
While gaining popularity among linguists, Wolof remains understudied,
particularly with regard to syntax and semantics. Recent works focusing on these areas,
such as Njie (1987), Robert (1991), Dunigan (1994), Nouguier-Voisin (2002), Russell
(2007), Torrence (2012), Martinovic (2013), and Dione (2013) have added great insight
into the language. The aim of this dissertation is to add to the study of Wolof with an in
depth treatment of its predicate argument structure. Treating all aspects of a language’s
argument structure is far too ambitious a goal for a single thesis so I will only attempt a
treatment of the argument structure of applicative predicates.
1.1. Clause Structure
There have been several syntactic treatments of Wolof in generative frameworks. Njie
(1982) focuses on the clause structure of Gambian Wolof. Dunigan (1994) and Russell
!4
(2007) describe and analyze Dakar Wolof focusing on clause structure. Torrence (2012)
and Martinovic (2013) explore the left periphery of Wolof clauses. Descriptive works on
Senegalese Wolof include Ngom (2003) and Fal (1999). Dione (2013) treats applicative
and causatives in Wolof using Lexical Functional Grammar. The rest of this section is
devoted to a brief description of Wolof. First, the basic clause structures are presented in
1.1.1. Nouns are described in 1.1.2., followed by adjectival modification in 1.1.3. The last
area of Wolof to be presented is focus and question formation in 1.1.4. Much of the
information given here is cited from Torrence (2012). Additional sources are cited when
used.
1.1.1. Clause Types
Neutral sentences in Wolof are SVO. As for verbal agreement, the verb agrees with the
subject in person and number. Agreement is not marked on the verb as affixes, rather
person and number are indicated on a subject marker that is considered an independent
word. The linear order of subject markers in relation to the verb depends on the tense,
aspect, mood, negation, and clause type. A list of clause types and agreement is giving in
Table 1. There is no object agreement in Wolof.
Table 1: Wolof Clause Types
(cited from Torrence 2012:29-31)
Type
Example
Use
-Na Clause
a. Xale yi
lekk-na-ñu gato bi.
child the.PL eat-FIN-3PL cake the
‘The children ate the cake.’
The entire clause is
new information.
No subconstituent
is in focus.
Negative
b. Xale yi
lekk-u-ñu gato bi.
child the.PL eat-NEG-3PL cake the
‘The children did not eat the cake.’
No emphasis on
anything. Negative
of na-clause
Subject
Cleft 1
c. Xale yi
a lekk gato bi.
child the.PL COP eat cake the
‘It’s the children who ate the cake.’
Subject in focus
!5
Subject
Cleft 2
d. Xale yi
ñu a lekk gato bi.
child the.PL 3PL COP eat cake the
‘It’s the children who ate the cake.’
Subject in focus
Negative
Subject
Cleft 1
e. Xale yi
a lekk-ul gato bi.
child the.PL COP eat-NEG cake the
‘It’s not the children who ate the cake.’
Negative of subject
cleft
Negative
Subject
Cleft 2
f. D-u
Negative of subject
cleft
Non-subject
Cleft
g. Gato bi l-a
xale yi
lekk.
cake the XPL-COP child the.PL eat
‘It’s the cake that the children ate.’
Non-subject in
focus
Subjunctive
h. Bëgg-na-a
ñu lekk-ko.
want-FIN-1SG 3PL eat-3SG
‘I want them to eat it.’
CP complement of
predicates of
desire, command,
wish, etc.
Adverbial
i. Tusuur ñu lekk-ko.
always 3PL eat-3SG
‘They always eat it.’
CP/TPs that are
introduced by
certain adverbs in
the left periphery
Optative
j. Xale yi
nañu
lekk gato bi!
child the.PL OPT-3PL eat cake the
‘The children, may they eat the cake!’
Wish or desire of
speaker
Negative
Optative
k. Xale yi
b-u ñu
lekk gato bi!
child the.PL C-NEG-3PL eat cake the
‘The children, may they not eat cake!’
Wish of desire of
speaker
Progressive
l. Xale y-àng-i
lekk gato bi.
child CL-PROG-LOC eat cake the
‘The children are eating the cake.’
Ongoing actions or
current states
Subject Focus
Progressive
m. Xale y-àng-ii
di
lekk gato bi.
child CL-PROG-LOC IMPERF eat cake the
‘It’s the children who are eating the cake.’
Subject is in focus
with ongoing
actions or current
states
xale yi
a lekk gato bi.
IMPERF-NEG chid the.PL COP eat cake the
‘It’s not the children who ate the cake.’
!6
Non-Subject
Focus
Progressive
n. Gato b-àng-ii
xale yi
di
lekk. Non-subject in
cake CL-PROG-LOC child the.PL IMPERF eat
focus
‘It’s the cake that the children are eating.’
Predicate
Focus
Progressive
o. Xale yi da-ñu lekk gato bi.
child the.PL do-3PL eat cake the
‘The children did eat the cake.’
‘Eat the cake is what the children did.’
Focus on predicate
or predicate (and
complement) of a
clause, explanation
Modal
p. Ma togg-al-la ceeb bi?
1SG cook-BEN-2SG rice the
‘Should I cook you the rice?’
Request
Exclamative
q. Aka mu leen dóór!
EXCL 3SG 3PL hit
‘How he hit them!’
Exclamations
A detailed description of each of these clause types is not needed to understand
the data in the following chapters. I would like to take a closer look, however, at several
of these clause types because their relevance in chapters 3 and 4. The first clause type is
what Torrence calls the -na clause. Here the entire clause expresses new information. I
refer to this clause as a neutral clause rather than a -na clause. As can be seen the lexical
subject xale yi ‘the children’ precedes the verb, lekk ‘to eat’ and the subject markers
(bolded) appear after the verb and agree with the lexical subject for person and number.
Neutral clause (repeated from a. in Table 1)
(1)
Xale
yi
lekk-na-ñu gato
child the.PL eat-FIN-3PL cake
‘The children ate the cake.’
bi.
the
I note here that the orthography I use represents the subject markers as independent
words.1 The clause in (1) from Torrence is given in (2) using the orthography adopted
here (see Fal 1999 and Diouf 2003 for more on orthography).
1
The bound/independent status of inflectional morphemes does not affect the analysis offered and spelling
conventions do not offer insight to structure. The inflectional morphemes are clitics and are dependent
phonetically on the verb. I mention the difference only to avoid confusion.
!7
(2)
Xale yi
lekk nañu gato
child DEF1 eat
3PL
cake
‘The children ate the cake.’
bi.
DEF1
One final note on the Wolof examples presented in this thesis is that I do not break down
subject markers morphologically in the examples. I present them as a single unit and
gloss them simply by the person and number features and the semantic features such as
predicate focus or object focus, as seen in (2). This is not motivated by theoretical
concerns but is done for convenience.
The next clause type that will be important in this thesis is a cleft structure that is
used to focus the subject. A copular element a follows the clefted subject xale yi ‘the
children’.
Subject cleft 1 (repeated from c. in Table 1)
(3)
Xale
yi
a
lekk
gato
child
the.PL COP1 eat
cake
‘It’s the children who ate the cake.’
bi.
the
A second way of forming a subject cleft is to include a subject marker that agrees
with the subject for person and number. In (4), the particle ñu reiterates the subject xale yi
‘the children’ and precedes the particle a. While not shown in Table 1, the vowel contact
between ñu+a results in the form ñoo (see Ka 1987 for more on morphophonological
patterns in Wolof).
Subject cleft 2 (repeated from d. in Table 1)
(4)
Xale yi
ñu
a (>ñoo) lekk
child the.PL 3PL COP1
eat
‘It’s the children who ate the cake.’
gato
cake
bi.
the
The non-subject cleft is used to focus syntactic objects. The focused object, gato
bi ‘the cake’, appears in the sentence initial position followed by the particle la. The
!8
subject, xale yi ‘the children’, follows the cleft particle and the verb, lekk ‘to eat’, follows
the subject.
Non-subject cleft (repeated from g. in Table 1)
(5)
Gato
bi
l-a
cake
the
XPL-COP
‘It’s the cake that the children ate.’
xale
child
yi
the.PL
lekk.
eat
Progressive clauses involve a progressive morpheme in the subject marker. In (6),
the y- represents the class of the subject nominal, in this case plural. It combines with a
progressive morpheme, àng, and finally a locative morpheme -i which indicates the
proximity of the subject.
Progressive with lexical subject (repeated from l. in Table 1)
(6)
a.
Xale
y-àng-i
lekk gato bi.
child CL-PROG-LOC eat
cake the
‘The children are eating the cake.’
Another possibility is that yàngi is both progressive and subject focus as proposed by
Zribi-Hertz and Diagne (2002:839). They argue the -a particle marks subject focus while
-ng is the copular particle. Like Torrence, they assume -i is a locative morpheme involved
in obviation.
(7)
y(i>a)-a
DEF.PL-SFOC
-ng
-i
COP1 LOC1
As mentioned earlier, the morphological division of subject markers is not crucial to a
discussion of Wolof applicatives, which are the subject matter of this thesis. I adopt
Torrence’s assertion that -a is the copular particle.
!9
A second way to express ongoing actions or current states is to use both a lexical
subject, xale yi ‘the children’, and a corresponding subject marker, ñu ngi as in (6).2 This
second type of progressive clause is commonly used when the subject is not focused and
is not new information. In the following example, I assume -ng as a progressive
morpheme.
Progressive
(8)
Xale yi
ñu
ng-i-y
lekk
child DEF1 3PL PROG-LOC-IMP eat
‘The children are eating the cake.’
gato
cake
bi.
DEF1
I assume as well that -i is a locative morpheme that marks the proximity of the subject
since the marker can also surface as ng-a when the subject is not proximal to the speaker.
In (8) the children are close to the speaker or have been mentioned recently in the
conversation while in (9) the children are not near the speaker or were mentioned further
back in the conversation.
Progressive
(9)
Xale yi
ñu
ng-a-y
lekk
child DEF1 3PL PROG-LOC-IMP eat
‘The children are eating the cake.’
gato
cake
bi.
DEF1
The last clause type I will talk about is what Torrence calls the predicate focus
progressive. I refer to it simply as predicate focus. In this clause, the subject marker is
made up of two morphemes, da- and the corresponding agreement for person and number
of the subject, -ñu in the case of (10). The subject marker precedes the lexical verb, lekk
‘to eat’.
2
Dunigan (1994) argues that in these cases, the lexical subject is in a topic position and the subject marker
is in the subject position of the structure.
!10
Predicate focus (repeated from o. in Table 1)
(10) Xale
yi
da-ñu
lekk
gato
bi.
child
the.PL
do-3PL
eat
cake
the
‘The children did eat the cake. / Eat the cake is what the children did.’
A final note about Wolof clauses is that Wolof does not have a syntactic passive
construction (Dione 2013). Sentences with a passive interpretation are expressed using a
3SG impersonal pronoun in subject position. A more detailed description and analysis of
the different clause types are given in Torrence (2012) and Zribi-Hertz and Diagne (2002)
(see also Njie 1982, Dunigan 1994, Fal 1999, Ngom 2003, and Russell 2007 for more on
Wolof clause structure).
1.1.2. Nouns and DPs
Wolof nouns are divided into nominal classes. Torrence (2012) identifies fifteen noun
classes, which he classifies into eight singular, two plural, two locative, and three
defective classes. Noun class is not marked on the noun itself but on the determiner. The
determiner is made up of a consonant that reflects the nominal class and a vowel that
expresses relative location of the noun. The singular classifiers are: b-, g-, l-, j-, w-, s-,
m-, k-, and the plural classifiers are: y-, ñ-. The three defective classifiers are f- (locative),
n- (manner), c- (preposition).
!11
Table 2: Nominal classifiers
Singular
Plural
(Torrence 2012: 16)
xaj bi
‘the dog’
bi-class
gaal gi
‘the boat’
gi-class
ndap li
‘the pot’
li-class
wax ji
‘the talk’
ji-class
jën wi
‘the fish’
wi-class
ndaw si
'the young woman’
si-class
saw mi
'the urine’
mi-class
nit ki
'the person’
ki-class
ja yi
‘the markets’
yi-class
góór ñi
'the men’
ñi-class
When the noun is definite, these classifiers combine with either -i or -a. When -i
is used, it indicates proximity or salience of the noun, be it conversational, physical, or
temporal. When -a is used it indicates the distance or non-salience of the noun.
(11)
a.
xaj
b-i
dog CL-PROX
‘the dog mentioned recently (in the conversation)’
‘the dog that is close (physically)’
‘the dog that existed recently/[currently exists]’
b.
xaj b-a
dog CL-DISTAL
‘the dog mentioned a while ago (in the conversation)’
‘the dog that is far away (physically)’
‘the dog that existed a long time ago’
(Torrence 2012:17-18)
Indefinite determiners consist of the vowel a- and the nominal classifier.
!12
(12)
a-b
xaj
a-CL dog
‘a dog’
(Torrence 2012:18)
Indefinite nouns can also be expressed as a bare noun or with the numeral benn ‘one’.
(13)
a.
xaj
dog
‘a dog’
b.
benn xaj
one
dog
‘a dog’
(Torrence 2012:18)
The defective classifiers are used only with demonstratives and pronouns.
Table 3: Defective Classifiers (Torrence 2012: 16)
Class
Demonstrative
Wh-pronoun
fi-class ‘locative’
foofu
‘aforementioned place’
fan
‘where?’
ci-class ‘prepositional’
coocu
‘in/at/on aforementioned place’
can
‘in/at/on where?’
ni-class ‘manner’
noonu
‘aforementioned way’
nan
‘how, in what way?’
The order of constituents within the DP is as follows.
Table 4
INDEFINITE DETERMINER
NUMERAL
POSSESSIVE PRONOUN
SIMPLEX QUANTIFIER
noun
DEFINITE DETERMINER
DEMONSTRATIVE
CREL + complement
COMPLEX QUANTIFIER
!13
1.1.3. Adjectives
Adjective is not a grammatical category in Wolof (McClaughlin 2004, Tamba et al.
2012). Qualitative adjectives are expressed by a relative clause containing a stative verb
that describes the property in question.
(14)
(15)
a.
bal
bu
mag
ball CREL
be.big
‘a big ball’ (lit: a ball that is big)
b.
bal
bi
mag
ball
CREL be.big
‘the big ball’ (lit: the ball that is big)
a.
bal
bu
tutti
ball CREL be.small
‘a small ball’ (lit: a ball that is small)
b.
bal
bi
tutti
ball
CREL be.small
‘the small ball’ (lit: the ball that is small)
The (a) examples in (14) and (15) show a relative clause modifying an indefinite noun
while the (b) examples involve a definite noun. The complementizer in definite relative
clauses is homophonous with the definite determiner. The exact mechanics of relative
clauses in Wolof is not crucial here, so I will not look at it further. For more information
on relative clauses and adjectival modification in Wolof see Torrence (2012), McLaughlin
(2004).
1.1.4. Focus and Question Formation
Focus and wh-question formation in Wolof involve the cleft structures mentioned earlier
(see Table 1 and examples 4 and 5) (Torrence 2012, Martinovich 2013). The focused or
!14
questioned element (underlined) is placed at the beginning of the sentence followed by
the appropriate subject marker and a copular particle, -a.
Subject cleft
(16) a.
Ayda
moo (>mu a) lekk
Ayda
3SG.COP1
eat
‘It is Ayda who ate dibi.’
dibi.
dibi
Non-subject cleft
b.
Dibi
l-a
Ayda
dibi
XPL-COP
Ayda
‘It is dibi that Ayda ate.’
lekk.
eat
Subject wh-question
(17) a.
Kan moo (>mu a) lekk
who 3SG.COP
eat
‘Who ate dibi?’
dibi?
dibi
Non-subject wh-question
b.
Lan
l-a
Ayda
what XPL-COP Ayda
‘What did Ayda eat?’
lekk?
eat
Wh-questions are types of focus constructions. I assume the focused NP/wh-phrase, if
present, is found in CP or focus position in the left periphery. For detailed descriptions
and analyses of focus and question formation see Torrence (2012) and Martinovich
(2013).
1.2. Methodology
The majority of the data presented in this thesis are new data collected in the field in
Saint Louis, Senegal. Only first language Wolof speakers who had grown up in Saint
Louis region were consulted for data collection to control for dialectal variation. A total
!15
of seven consultants responded to grammaticality judgement tasks, elicitation tasks, and
acceptability tasks. All data were verified by at least three different speakers.
Data were collected in the city of Saint Louis, Senegal, situated on the northwest coast of
Senegal just south of the border with Mauritania. The data were collected during two
research trips, totaling five months in Saint Louis during the spring and fall of 2013.
None of the consultants have any formal instruction in Wolof. Five of the seven
consultants had formal instruction in French through at least a high school level of
education. All consultants are Wolof and French bilingual speakers. The consultants
ranged in age from twenty to mid-sixties and were all male.
Data were collected using French as the metalanguage. Three types of tests were
preformed in order to target different information. Elicitation tasks, such as translations,
were used to define basic utterances and identify relevant vocabulary and grammatical
information to facilitate further testing. Once baseline sentences and structures were
established with elicitations, negative evidence was collected using grammaticality
judgements and acceptability judgements. Grammaticality judgements were used to target
syntactic constraints. Limited context was provided when presenting sentences and the
consultants were asked to judge sentences.
Grammaticality judgements show what structures are acceptable but they do not
offer much insight into the semantic information encoded in a sentence (Matthewson
2004). To target the semantics of tests sentences, acceptability tasks were used. The
acceptability tasks are similar to grammaticality judgements except that the sentence in
question is presented in a detailed context that selects for a particular semantic property
or interpretation. The consultant was then asked to rate whether the sentence was
appropriate within the provided context. Contexts were presented as verbal descriptions,
stories, pictures, cartoon drawings, or acting.
!16
1.3. Overview of Thesis
The rest of the thesis is organized as follows. Chapter 2 offers a general description of
applicative structures and gives an overview of some of the major analytical approaches
for applicatives. Previous research published on Wolof applicatives will be presented at
the end of chapter 2. Chapter 3 focuses on the description and analysis of benefactive and
dative applicatives. Instrumental and locative applicatives and their analysis are the
subject of chapter 4. Concluding remarks and remaining questions are found in chapter 5.
!17
Chapter 2
2. APPLICATIVES: THEORETICAL BACKGROUND
Applicatives are of interest to linguists because of the particular challenges they pose for
syntactic theory. Early formal approaches to applicative sentences attempted to elucidate
a single structure from which all types of applicatives arise (Baker 1988, 1992, Larson
1988, Marantz 1993). More recent proposals (McGinnis 2005, Pylkkänen 2008, Georgala
2012) involve multiple structures for different types of applicatives. Such approaches
have many advantages over more traditional, single structure approaches, but still fail to
account for the full range of applicative variation exhibited. I maintain that a subtler
approach is needed, involving unique structures for different types of applicatives, based
on a detailed exploration and analysis of Wolof applicatives. The need for different
structures casts doubt on treating applicatives as a construction because although they
share descriptive similarities (adding an argument), structurally they are quite different. I
assume that “applicative” refers to a class of descriptively similar structures. Note that
this assumption fits within the Minimalist approach to syntactic theory, where we have
moved away from constructions in an attempt to find the primitive components of the
grammar.
In this chapter, I present an overview of applicatives and their properties using
examples from various languages. Applicatives are grouped into types based on the theta
role of the applied object (e.g. benefactive, dative, instrument, locative, reason, etc.).
Cross-linguistically, these different applicatives tend to show common patterns of
behaviour in relation to object properties and syntactic configuration. Differences in
behaviour are also discussed and provide motivation for classifying applicatives by type
rather than assuming a single applicative construction. I then turn to theoretical work on
applicatives and then review several previous analyses of Wolof applicatives. Building
off these analyses, in the last section, I show that a subtler approach is needed and
!18
propose different structures for the different applicatives found in Wolof. These structures
will be motivated and discussed in more detail in chapters 3 and 4.
2.1. Applicatives
The term applicative refers to verb forms1 in which the verbal valency has been increased
by one.2 Payne (1997) defines an applicative as “... a valence increasing operation that
brings a peripheral participant onto center stage by making it into a direct object. The
‘new’ direct object is sometimes referred to as the applied object (186).” This is
illustrated in English in examples (1-3) below. The first sentence shows an example of the
transitive, non-applicative, use of write. The sentence in (2) shows an applicative in
English with the same verb. The sentence now contains two objects, Sam, the recipient,
and a letter. The sentence in (3) has two complements like the applicative version in (2)
Sam and a letter, but the recipient is contained in a prepositional phrase. I call examples
like (3) prepositional complement sentences.
(1)
Bill wrote a letter.
(2)
Bill wrote Sam a letter.
(3)
Bill wrote a letter to Sam.
I note here that the term ‘applicative’ is at times confused with the term ‘double
object construction’. While the two are sometimes used interchangeably, there is an
important distinction between them. A double object construction involves two direct
objects associated with a single predicate, as seen in (4). It is also an applicative by
definition since the valency of the verb bake has increased by one from “Bill baked a
cake.”.
1
‘Applicative’ is also used to refer to sentences that contain an applicative verb. Note that ‘verb forms’
does not necessarily mean that applicative forms are morphological distinct from non-applicative forms. In
some languages, like English, there is no overt change in verb form.
2
I exclude causative constructions from the discussion of applicatives, even though they technically fall
under this definition, as they involve the addition of an external argument, not an internal one.
!19
(4)
Bill baked Sarah a cake.
The valency increase of an intransitive verb, however, does not lead to a sentence with
two objects so it can’t be called a double object construction. The term applicative still
correctly qualifies the sentence in (5b) and others like it, as a valency increase has
occurred between (5a) and (5b) even though there is only one object present, Katonga.
Luganda
(5)
a.
b.
Mukasa ya-tambu-dde
Mukasa 3SG.PST-walk-PST
‘Mukasa walked.’
Mukasa ya-tambu-le-dde
Mukasa 3SG.PST-walk-APPL-PST
‘Mukasa walked for Katonga.’
Katonga.
Katonga
(Pylkkanen 2008:20)
In other words, double object constructions are applicatives, but not all
applicatives are double object constructions. Note that I consider lexically ditransitive
verbs, like give, as applicatives. This is not an obvious conclusion since the valency of
the verb has not increased. Ditransitive verbs, like give, cannot be used intransitively, and
strongly resist even transitive usage. Nevertheless, the second half of Payne’s definition
does describe ditransitive verbs in double object constructions because the indirect object
has been metaphorically moved from a peripheral position within the PP, (6a), to a direct
object, (6b).
(6)
a.
I give presents to the children.
b.
I give the children presents.
It is important to note that I do not assume that the sentence in (b) is derived from (a)
unlike Larson proposes (1988). They involve different derivations and structures
following Marantz (1993) and Harley (2002). Predicates like these which involve a nonmorphologically marked applied object are called valency-preserving (Creissels 2004).
!20
Now that applicative sentences have been briefly described, I turn to a discussion
of generally recognized properties of applicatives.
2.1.1. Typology Cross-linguistically, applicatives are found in languages from all over the world. They
are common in three geographical areas: Africa, the western Pacific region, and the
Americas according to the World Atlas of Language Structures Online (Polinsky 2013).
There is quite a bit of variation within applicative sentences. There are multiple ‘flavors’
of applicatives based on the thematic role of the applied object, such as benefactive, goal,
instrumental, locative, and reason. The inventory of which applicatives are available
depends on the language in question. In some languages, applicatives are very productive
and can express multiple theta roles, like Bantu languages, while in others, like English,
applicatives are more constrained.
In English, applicatives have limited productivity: benefactive or recipient/source
arguments can be expressed as a direct object but only with a certain class of verbs, those
which convey a transfer of possession, like give, send, sell, write. The sentence in (7a)
shows an applicative with a recipient applied object, Sam, and (7b) shows a benefactive
object, Sarah. Note that in both sentences the possession of the theme objects, a letter
and a cake, is transferred to the applied object.
(7)
a.
Bill wrote Sam a letter.
b.
Bill baked Sarah a cake.
Since both applicatives express a transfer of possession, I call them dative applicatives.3
Examples of dative applicatives from other languages are given in (8) - (12). Greek,
Basque, and Albanian examples are similar to English in that there is no overt applicative
marker attached to the verb.
3
This type of applicative has been called ditransitive or possessor dative, although the exact terminology
varies widely.
!21
Greek
(8)
O
Nikos
edhose
tis
the.NOM Nick.NOM gave.3SG the.GEN
‘Nick gave Mary a book.’
Basque
(9)
Jonek Norari liburua
ekarri
J-ERG N-DAT book-ABS bring
‘Jon brought Nora the book.’
Marias
Mary.GEN
ena
a.ACC
vivlio.
book.ACC
(Georgala 2012:104)
dio.
AUX1
(Oyharçabal 2010:235)
Albanian
(10) Agimi
i mban Drites
çanten time.
Agim.NOM CL holds Drita.DAT bag.ACC my
‘Agim holds my bag for Drita.’
Kinyarwanda
(11) Umugóre y-iim-ye
ábáana ibíryo.
woman
she-refuse-ASP children food
‘The woman refused food to the children.’
(Pylkkänen 2008:21)
(Kimenyi 1980:31)
Some languages, like Mandarin, on the other hand, require an additional
morpheme, like gĕi, in applicative sentences. If gĕi is missing, the sentence is
ungrammatical (12b).
Mandarin
(12) a.
b.
Wŏ mài-gĕi-le
Măli yī-ge shŏubiăo.
1SG sell-GEI-PERF Mali 1-CL watch
‘I sold Mali a watch.’ (cited from Georgala 2012)
*Wŏ mài-le
Măli
1SG sell-PERF
Mali
‘I sold Mali a watch.’
yī-ge
1-CL
shŏubiăo.
watch
(Paul and Whitman 2010: 264)
!22
While in English only dative applicatives are possible, in other languages,
applicative sentences can express a wide range of theta roles as seen in Chaga (13) and
Ndendeule (14).
Benefactive applicative - Chaga
(13) a.
N-a-i-lyi-i-a
m-ka
FOC-SP-prs-eat-APPL-FV wife
‘He is eating food for his wife.’
k-elya.
food
Instrumental applicative - Chaga
b.
N-a-i-lyi-i-a
ma-woko
FOC-SP-prs-eat-APPL-FV
hand
‘She is eating food with her hands.’
k-elya.
food
(Marantz 1993:121-122)
Malefactive applicative - Ndendeule
(14) a.
ma-yani γa-ki-βa-yͻmͻ-εl-a
ma-chi βa-lumba.
6-baboon 6SA-PST-break-APP-FV 6-water 2-hunter
‘The baboons finished the hunter’s water.’
Locative applicative - Ndendeule
b.
βa-lumba βa-ki-tul-il-a
nyama
pa-manyahi.
2-hunter 2-PST-skin-APP-FV 9.animal 16-grass
‘The hunters skinned the animal on the grass.’
Reason applicative - Ndendeule
c.
m-wana a-ki-lel-el-a
ki-hembe.
1-child 1SA-PST-cry-APP-FV 7-knife
‘The child cried for a knife.’
(Ngonyani 1996:18)
Some languages allow an applied object with intransitive verbs, like
Kinyarwanda. A locative applied object, intebe ‘chair’ has been added to the intransitive
!23
verb iica ‘sit’ in (15). A benefactive object has been applied to the intransitive verb kor
‘work’ in (16).
Locative applicative - Kinyarwanda
(15) Umugabo y-iica-yé-ho
íntebe.
man
he-sit-ASP-on chair
‘The man is sitting on the chair.’
Benefactive applicative - Kinyarwanda
(16) Umugóre a-rá-kor-er-a
woman
she-PRES-work-APPL-ASP
‘The woman is working for the man.’
(Kimenyi 1980:38)
umugabo.
man
(Kimenyi 1980:32)
In some languages, applicatives can be iterated. The sentence in (17) has a
locative applied object, intebe ‘chair’, and a benefactive applied object, umugabo ‘man’.
Additionally, there are two applicative suffixes attached to the verb, -i for the benefactive
and -ho for the locative.
Locative and benefactive applicative - Kinyarwanda
(17) Úmwáana y-iicar-i-yé-ho
íntebe umugabo.
child
he-sit-APPL-ASP-LOC chair
man
‘The child is sitting on the chair for the man.’
(Kimenyi 1980:113)
In (18), there are three objects, a benefactive object umugóre ‘woman’, a dative
object ábáana ‘children’, and a theme object ibíryo ‘food’. Only one applicative
morpheme appears on the verb and it is associated with the presence of the benefactive
object.
Triple object construction - Kinyarwanda
(18) Umukoôbwa a-rá-hé-er-a
girl
she-PRES-read-APPL-ASP
umugóre
woman
ábáana ibíryo.
children food
‘The girl is giving food to the children for the woman.’
(Kimenyi 1980:32)
!24
Predicates with dative objects in Kinyarwanda don’t show overt morphology, similar to
English, Greek, and Albanian applicatives. Recall from section 2.1. that I assume double
object constructions with or without overt applicative morphology are applicatives so the
sentence in (18) involves two applicative heads, dative and benefactive, only one of
which is overtly realized.
From this brief survey, we see applicatives can express a variety of theta roles
including dative (recipient or source), benefactive, instrumental, locative, and reason.
They can be signaled by overt or null morphology and they are compatible with transitive
and intransitive predicates. In the next section, I focus on the object properties of
applicatives.
2.1.2. Object properties
Object properties are often a starting point for applicative research. Researchers attempt
to determine to what extent the objects in applicative sentences show similar syntactic
properties as “normal” direct objects (e.g. of a monotransitive verb). Common object
properties are passivization, pronominalization, and adjacency to the verb. In applicatives
that have two objects, one or both objects can have object properties depending on the
language and applicative type in question. When both objects display object properties,
the applicative is classified as symmetrical. When only one object has direct object
properties, the applicative is classified as asymmetrical. In the literature, people often
refer to the languages themselves as symmetrical or asymmetrical (see Bresnan and
Moshi 1990, Ngonyani 1996, 1998) but looking at the full range of data this classification
is not entirely accurate. As will be seen in this section, some applicative types in certain
languages show symmetrical properties while other types in the same language show
asymmetrical properties. This mixed behaviour is particularly relevant to the discussion
of Wolof applicatives in chapters 3 and 4.
!25
2.1.2.1. Passivization
One test for objecthood is passivization. If an argument can be made the subject of a
passive sentence, it behaves like a direct object. In symmetrical applicatives, either object
can be the subject of a passive while in asymmetrical ones only one object can be.
English, for example, only allows the applied object, Jill, to become the subject of a
passive sentence (19b). The direct object, a cake, cannot be the subject of a passive
applicative sentence, (19c); it loses its object properties when the applied object is
present.
(19)
a.
Mary baked Jill a cake.
b.
Jill was baked a cake (by Mary).
c.
*A cake was baked Jill (by Mary).4
Compare this to the prepositional complement sentence in which the theme, a cake, can
be the subject of a passive and the benefactive object, Jill, cannot.
(20)
a.
A cake was baked for Jill (by Mary).
b.
*Jill was baked a cake for (by Mary).
In (19), we saw in English that the applied object acts like a direct object rather than
oblique objects, while the original direct object, the theme, no longer exhibits object
properties as it does in the prepositional complement sentence in (20).
In some languages, however, both objects in applicatives can be the subject of a
passive sentence. Kinyarwanda has symmetrical applicatives that allow either object to be
the subject of a passive sentence. Example (21a) involves two applied objects, a
benefactive umugabo ‘man’ and a dative ímbwa ‘dog’, resulting in a total of three direct
objects. Symmetry is shown because the subject position can be filled by any one of the
three objects in a passive, as shown in (b-d).
4
In certain dialects of English, theme passivization is acceptable.
!26
Active sentence
(21) a.
Umugóre a-rá-hé-er-a
umugabo ímbwa
woman
she-PRES-give-APPL-ASP man
dog
‘The woman is giving food to the dog for the man.’
ibíryo.
food
Passive - Benefactive subject
b.
Umugabo a-rá-hé-er-w-a
ímbwa ibíryo n’ûmugóre.
man
she-PRES-give-APPL-PASS-ASP dog
food by.woman
lit: ’The man is given food to the dog by the woman.’ (The man benefits
from the woman giving food to the dog.”
Passive - Dative subject
c.
ímbwa i-rá-hé-er-w-a
umugabo ibíryo
dog
she-PRES-give-APPL-ASP man
food
‘The dog is given food for the man by the woman.’
n’ûmugore.
by.woman
Passive - Theme subject
d.
ibíryo bi-rá-hé-er-w-a
umugabo ímbwa n’ûmugore.
food
she-PRES-give-APPL-ASP man
dog
by.woman
‘The food is given to the dog for the man by the woman.’
(Kimenyi 1980: 65-66)
As mentioned earlier, I do not classify Kinyarwanda as a symmetrical language
because there is variation based on the type of applicative sentence. Locative applicatives
do not show symmetrical behaviour like the benefactive applicative in (21). In locative
applicatives, only the applied object can raise to subject position (22a). The theme cannot
be the subject of a passive sentence (22b).
Passive - locative subject
(22) a.
Umwaana y-a-menn-w-e-ho
amaazi n’umubooyi.
child
SP-PST-pour-PASS-ASP-APPL water
by.cook
‘The water was poured on the child by the cook.’
!27
Passive - theme subject
b.
*Amaazi y-a-menn-w-e-ho
umwaana n’umubooyi.
water
SP-PST-pour-PASS-ASP-APPL
child
by.cook
‘The water was poured on the child by the cook.’
(Zeller and Ngoboka 2006:102)
For this reason, I classify applicative types based on their symmetrical or
asymmetrical properties, rather than classifying them by languages. In English, which
only have one type of applicative, they are asymmetrical. In Kinyarwanda, benefactive
and dative applicatives are symmetrical while locative applicatives are asymmetrical.
2.1.2.2. Pronominalization
The second object property under consideration is the pronominalization of objects using
a verbal clitic or affix. In Bantu languages, like the Kinyarwanda example in (23),
pronominalization is generally called ‘object marking’ and involves a prefix (underlined)
attached to the verb. The pronominal prefix matches the noun it replaces for noun class.
(23)
a.
Umugóre a-rá-bi-he-er-a
umugabo
woman
she-PRES-it-give-APPL-ASP man
‘The woman is giving it to the dog for the man.’
ímbwa
dog
b.
Umugóre a-rá-yi-he-er-a
umugabo
woman
she-PRES-it-give-APPL-ASP man
‘The woman is giving food to it for the man.’
ibíryo.
food
c.
Umugóre a-rá-mu-he-er-a
ímbwa
woman
she-PRES-him-give-APPL-ASP dog
‘The woman is giving food to the dog for him.’
ibíryo.
food
(Kimenyi 1980:66)
!28
Symmetrical applicatives allow both, or all, the objects to be pronominalized
using verbal prefixes, as in (23), while asymmetrical applicatives only allow one. For
example, in Kiswahili, the benefactive object can be marked but not the theme object.
Kiswahili - Benefactive object agreement
(24) a.
Juma a-li-wa-let-e-a
wa-toto
mw-alimu.
Juma 1-PST-1-bring-APP-FV 2-child
1-teacher
‘Juma brought the teacher for the children.’
(Ngonyani 1998:84)
Kiswahili - Theme object agreement
b.
*Juma a-li-m-let-e-a
mw-alimu wa-toto.
Juma 1-PST-1-bring-APP-FV 1-teacher 2-child
‘Juma [brought] the teacher for the children.’
(Ngonyani 1998:86)
As with passivization, pronominalization shows that Kinyarwanda locative applicatives
are asymmetrical. The applied object can be marked on the verb but not the theme.
Kinyarwanda - locative object marking
(25) a.
Umubooyi y-a-mu-menn-ye-ho
amaazi.
cook
SP-PST-OM-pour-ASP-APPL water
‘The cook poured water on him/her.’
Kinyarwanda - theme object marking
b.
*Umubooyi y-a-ya-menn-ye-ho
umwaana.
cook
SP-PST-OM-pour-ASP-APPL child
‘The cook poured it on the child.’
(Zeller and Ngoboka 2006:102)
The pronominalization test will show that like Kinyarwanda, Wolof has both
symmetrical and asymmetrical applicatives (see chapters 3 and 4).
!29
2.1.2.3. Word Order
Asymmetrical applicatives, like those found in English and Kiswahili, only allow one
object, the dative/benefactive object, to appear adjacent to the verb. Julie cannot follow a
car as seen by the ungrammaticality of (26b). The same is true for the benefactive object,
ma-gazeti ‘papers’, in (27b).
(26)
a.
Sally sold Julie a car.
b.
*Sally sold a car Julie.
Kiswahili - Benefactive - theme order
(27) a.
Juma a-li-chor-e-a
ma-gazeti
Juma 1-PST-draw-APPL-FV 6-paper
‘Juma drew pictures for papers.’
picha.
10.picture
Kiswahili - Theme - benefactive order
b.
*Juma a-li-chor-e-a
picha
Juma
1-PST-draw-APPL-FV 10.picture
‘Juma drew pictures for papers.’
ma-gazeti.
6-paper
(Ngonyani 1998:81)
On the other hand, symmetrical applicatives, as illustrated in the examples below
from Kikuyu, allow either object to appear adjacent to the verb. The benefactive object in
(28) ciana ‘children’ can precede or follow the theme, mũbira ‘ball’.
Kikuyu - Benefactive > theme order
(28) a.
Mũ-geni a-ra-gũ-ĩ-ire
ci-ana
1-guest
1SA-PRG-buy-APPL-PF 8-child
‘The guest bought children a ball.’
mũ-bira.
3-ball
!30
Kikuyu - Theme > benefactive order
b.
Mũ-geni a-ra-gũ-ĩ-ire
mũ-bira ci-ana.
1-guest
1SA-PRG-buy-APPL-PF 3-ball
8-child
‘The guest bought a ball for the children.’
(Ngonyani and Githinji 2006:35)
In Kinyarwanda instrumental applicatives either the theme or instrument is allowed to be
adjacent to the verb as in Kikuyu but locative applicatives have fixed word order like
English and Kiswahili.
Instrumental applicative - theme > instrument order
(29) a.
Umugabo y-a-tem-eesh-eje
igiti
man
SP-PST-cut-APPL-ASP
tree
‘The man cut the tree with the machete.’
umuhoro.
machete
Instrumental applicative - instrument > theme order
b.
Umugabo y-a-tem-eesh-eje
umuhoro igiti.
man
SP-PST-cut-APPL-ASP
machete
tree
‘The man cut the tree with the machete.’
(Zeller and Ngoboka 2006:117)
Locative applicative - locative > theme order
(30) a.
Umubooyi y-a-menn-ye-ho
umwaana
cook
SP-PST-pour-ASP-APPL child
‘The cook poured water on the child.’
amaazi.
water
Locative applicative - theme > locative (Zeller and Ngoboka 2006:108)
b.
*Umubooyi y-a-menn-ye-ho
amaazi umwaana.
cook
SP-PST-pour-ASP-APPL
water
child
‘The cook poured water on the child.’
(Zeller and Ngoboka 2006:108)
!31
Word order shows that some applicatives in Kinyarwanda are symmetrical (e.g.
instrumentals) while others (e.g. locatives) are asymmetrical. A similar pattern will be
seen in Wolof applicatives in chapters 3 and 4.
2.1.2.4. Extraction
A third property, A-bar extraction, provides useful information about the objects in
applicatives. In English, the theme object what in (31a) is questioned, and the sentence is
acceptable. In (31b) the sentence’s acceptability is degraded when the dative object, who,
is questioned. Extraction thus shows the English applicative to be asymmetrical, just as
we have seen above.
(31)
a.
What did you give Julie?
b.
?Who did you give books?
In symmetrical applicatives, either object can be extracted. For example in Kiswahili, the
instrument can be questioned as in (32a) or the theme can be questioned like in (32b).
The wh-word, nini ‘what’, remains in situ but assuming wh-movement is required at LF
for interpretation (Huang 1982), these are still cases of A-bar extraction.
Instrument questioned - Kiswahili
(32) a.
wa-toto wa-li-vunj-i-a
nini ch-ungu?
2-child 2-PST-break-APP-FV what 7-pot
‘What did the children break the pot with?’
Theme questioned - Kiswahili
b.
wa-toto wa-li-vunj-i-a
nini
ma-we?
2-child 2-PST-break-APP-FV what 6-rock
‘What did the children break with the rocks.?’
(Ngonyani 1998:82)
As with the other object properties, an object’s accessibility for extraction
sometimes depends on the type of applicative sentence as seen in Kinyarwanda (33) and
!32
(34). Either object can be extracted in dative applicatives. In (33) the theme, igitabo
‘book’, has been relativized. In (34), the dative object, umukoôbwa ‘girl’ has been
relativized.
Kinyarwanda - Theme extracted
(33) N-a-boon-ye
igitaboi [umuhuûngu ya-a-haá-ye
umukoôbwa ti].
SP-PST-see-ASP book
boy
SP-PST-REL.give-ASP girl
‘I saw the book [that the boy gave to the girl].’
(McGinnis 2001:15)
Kinyarwanda - Dative extracted
(34)
N-a-boon-ye umukoôbwai [umuhuûngu y-a-haá-ye
I-PST-see-ASP girl
boy
ti igitabo].
he-PST-REL.give-ASP
‘I saw the girl to whom the boy gave the book.’
book
(Kimenyi 1980:68)
Kinyarwanda locative applicatives, however, do not allow the theme to be extracted, only
the locative object. The sentence in (35) is ungrammatical because the theme object,
igitabo ‘book’, has been relativized while relativizing the locative object in (36), ishuûri
‘school’, is grammatical.
Theme extracted from locative applicative - Kinyarwanda
(35)
*y-a-tw-eerets-e
igitaboi [úmwáalímu y-oóhere-jé-ho
ishuûri ti].
SP-PST-OP-show-ASP
book
school
teacher
SP-REL.send-ASP-APPL
‘He showed us the book [that the teacher sent to school].’ (McGinnis 2001:15)
Locative extracted
(36) Umugabo y-a-tw-eerets-e
ishuûrii [úmwáalímu y-oóhere-je-ho…
man
SP-PST-OP-show-ASP school teacher
SP-REL.send-ASP-APPL
ti
igitabo].
book
‘The man showed us the school to which the teacher sent the book.’
(Kimenyi 1980:95)
!33
The data once again indicate that both symmetrical and asymmetrical applicatives exist in
the same language. Languages are not necessarily symmetrical or asymmetrical. Rather
symmetrical or asymmetrical classification is better applied to applicative types
themselves.
2.1.3. C-command
A second issue for applicative research is the structural relationships between the objects.
Tests sensitive to c-command are often used for exploring structural relationships
between items. In applicatives, previous research has shown the objects have an
asymmetrical c-command relationship, where one of the objects c-commands the other.
Whether the applied object c-commands the theme or vice-versa depends on the language
and the type of applicative in question.
Looking at data from quantifier binding in English applicatives, a quantified
dative object, like every worker, can bind an anaphor within the theme (37a) but the
theme, every paycheck, cannot bind an applied dative (37b).
Quantifier Binding - Applicative
(37) a. I sent every workeri hisi paycheck.
b. *I sent itsi owner every paychecki.
Applied > Theme
In the prepositional complement construction (38), a quantified DO, every check, can
bind an anaphor in the PPIO (38a) but the PPIO, to every worker, cannot bind a DO (38b).
Quantifier Binding - Prepositional Complement Construction
(38) a. I sent every checki to itsi owner.
b. ??I sent hisi paycheck to every workeri.
Theme > Applied
Similar patterns obtain for other c-command tests (39-48), as shown by Barss and Lasnik
(1986) and Larson (1988) (examples taken from Larson 1988:336-338).
Anaphor Binding - Applicative
(39) a. I showed Mary herself.
Applied > Theme
!34
b. *I showed herself Mary.
Anaphor Binding - Prepositional Complement Construction
(40) a. I showed Mary to herself.
b. *I showed herself to Mary.
Theme > Applied
Weak Crossover - Applicative
(41) a. Which mani did you send hisi paycheck?
b. *Whosei pay did you send hisi mother?
Applied > Theme
Weak Crossover - Prepositional Complement Construction
(42) a. Which checki did you send to itsi owner?
b. *Which workeri did you send hisi check to?
Theme > Applied
Superiority - Applicative
(43) a. Who did you give which paycheck?
b. *Which paycheck did you give who?
Applied > Theme
Superiority - Prepositional Complement Construction
(44) a. Which check did you send to who?
b. *Whom did you send which check to?
Theme > Applied
Each...the other - Applicative
(45) a. I showed each man the other’s socks.
b. *I showed the other’s friend each man.
Applied > Theme
Each...the other - Prepositional Complement Construction
(46) a. I sent each boy to the other’s parents.
b. *I sent the other’s check to each boy.
Theme > Applied
Negative Polarity Items - Applicative
(47) a. I showed no one anything.
b. *I showed anyone nothing.
Applied > Theme
Negative Polarity Items - Prepositional Complement Construction
(48) a. I sent no presents to any of the children.
b. *I sent any of the packages to none of the children.
Theme > Applied
Not all of these tests are amenable to applicatives in other languages. Negative polarity
items are not found in all languages, for example. Not all languages show syntactic
!35
superiority effects. Quantifier binding is one test that lends itself well to other languages
(Marantz 1993) and is often used in applicative research. I present the relevant data
below.
Looking at a range of languages, benefactives and datives typically c-command
theme objects, as in the English examples. In Swahili, quantifier binding shows the
applied object asymmetrically c-commands the theme. The applied object kila mwandishi
‘each other’ binds the anaphor chake ‘his’ in the theme object, (49a). The quantified
theme in (48b), however, cannot bind the anaphor, wake ‘its’, in the applied object.
Quantifier binding - Swahili
(49) a.
Ni-li-m-som-e-a
[kila
SP-PST-OP-read-APPL-FV each
‘I read each author his book.’
b.
mwandishi]i kitabu
author
book
chakei.
his
*Ni-li-m-som-e-a
[kila kitabu]i mwandishi wakei.
SP-PST-OP-read-APPL-FV each book
author
its
‘I read for its author each book.’
(Marantz 1993:117)
Instrumental applicatives in Kinyarwanda, however, show that the applied object does not
always c-command the theme. A quantified theme object, buri muryango ‘each door’ in
(50a), can bind an anaphor in the applied instrumental object, rwáwo ‘its’, but not viceversa.
Quantifier binding - Kinyarwanda
(50) a.
N-a-fúngul-ish-ije
buri muryangoi
I-PST-open-APPL-ASP each door
‘I opened each doori with itsi key.
úrufunguzo rwáwoi.
key
its
!36
b.
*N-a-fúngul-ish-ije
umuryano wáyoi buri
rufunguzoi.
I-PST-open-APPL-ASP door
its
each key
‘I opened itsi door with each keyj/*i.’
(A. Rutayoberana, p.c., cited in McGinnis 2005:195)
C-command tests show the applied object usually c-command the theme object but there
are exceptions, like instrumental applicatives in Kinyarwanda.
2.1.4. Summary
Given the inter- and intralinguistic variation in applicative sentences, it is important to
identify which types of applicatives are present in a language (e.g. benefactive,
instrumental, locative, etc.), what the properties of the objects are (e.g. symmetrical or
asymmetrical), and which object c-commands the other. Applicatives do not always show
the same properties or c-command, even within the same language as illustrated by
Kinyarwanda.
2.2. Syntactic Approaches
The goal of any formal syntactic analysis of applicatives is to account for the presence of
an additional direct object, (e.g. its selection and Case), the objects’ properties (e.g.
extraction, passivization, object marking, and word order), and the structural relationship
between the objects (e.g. binding, wh-movement, superiority, negative polarity items,
reciprocalization and reflexivization). There are many proposals available in the literature
to account for applicatives and their properties. I start with a very brief overview of preMinimalist approaches to applicatives (traditionally called double object constructions)
before turning in more detail to more recent proposals.
2.2.1. Issues in applicative research
What is challenging about applicatives is the presence of an additional object. Starting
with a transitive verb, the resulting applicative will have two ‘direct objects’. Starting
with an intransitive verb, the applicative will have one direct object. Major questions for
!37
applicative analysis are where is this additional object in the structure and how does it get
there. Early generative approaches assumed a ternary structure in which both objects
were sisters to each other and the verb (Baker 1988).
(51)
However, a ternary branching structure is problematic for theoretical and empirical
reasons. First, developments in Minimalist syntactic theory do not support a ternary
branching structure analysis for any type of sentence. Following Kayne (1984), binary
branching structures were adopted by many as the only possible type of structure in
generative syntax. Binary branching structure prevents both objects from simultaneously
being sisters of the verb. One object would necessarily be higher than the other.
(52) a.
b.
Empirical evidence supports a structure in which the objects are not sisters. Ccommand data show an asymmetry between the two objects. The applied object typically
c-commands the theme object, as shown in section 2.1.3. The structure in (51) is ruled
out because it predicts mutual c-command between the objects and (52a) is ruled out
because it predicts the second object will c-command the first. In English, however, the
first object c-commands the second. As for the structure in (52b), the first object ccommands the second, which matches the empirical data but we will see that it is unable
to account for applicatives cross-linguistically.
!38
Another issue for applicatives is Case assignment for both objects. If one assumes
a transitive verb has only one case, the applicative is problematic since the applied object
would then violate the Case Filter. How does the second object receive case, then and
escape the Case Filter if the verb has only one Case to assign or in the case of intransitive
verbs, if it has no Case at all? The other option is to assume the verb has two Cases to
assign, but then the prepositional variant is a mystery. If the verb assigns two Cases, the
use of a preposition to provide Case for oblique arguments would be redundant.
Some researchers, like Baker (1988), argue verbs in some languages, like
Kinyarwanda, can assign structural Case to two objects. Since the verb can case license
both objects, no preposition is needed to give Case to the second object. Verbs in other
languages, like English can also assign more than one Case, but one is structural case and
the other is inherent case. Under such an account the verb in applicatives assigns
structural case to one object and inherent case to the other while the verb in prepositional
complement constructions only assigns structural Case to one object and the preposition
is inserted to save the second object from the Case Filter. Why the verb sometimes
assigns inherent case, resulting in double object constructions, and sometimes does not,
resulting in the prepositional complement, is unclear. As we will see below, Marantz
proposes an alternative approach involving an additional verbal head that can introduce
and license the applied object, which avoids arbitrary stipulations of Case assignment.
2.2.2. Marantz
Marantz (1993) is one of the first linguists to propose that applied objects are not
introduced by the lexical verb but by a light verb (what I will call a functional head)
called Appl. The motivation for this separate projection in the structure of applicatives is
to capture the difference in semantic affectedness between the applied and theme objects.
In his book on grammatical relations (1984), he shows the asymmetry between the
dependency of objects and subjects. Changing the object changes the event described as
seen in (53) (examples from Marantz 1984:25).
!39
(53)
a.
b.
c.
d.
e.
throw a baseball
throw support behind a candidate
throw a boxing match (i.e. take a dive)
throw a party
throw a fit
Marantz argues that the event designated by a predicate is defined in combination with
the theme object. Assuming the verb is alone in depicting an event, one would expect the
events in all of the sentences in (53) to be the same, which is not true. Throwing a
baseball involves propelling a baseball through the air by use of one’s arm, a very
different event from throwing a fit. Throwing a fit describes a person who is expressing
dissatisfaction in a conspicuous manner. These examples show event meaning is derived
compositionally. Marantz claims that a verb, such as throw, take, or kill, is underspecified
as to the event described until it merges with the object.
The same cannot be said of subjects. Changing the subject has little effect on the
semantic role assigned to the object, as can be seen in (54). Unlike (53), the ‘NP’ in each
sentence has the same semantic role even though the subject has changed. In each
example, ‘NP’ is being propelled through the air by the subject (examples from Marantz
1984:26).
(54)
a.
The policeman threw NP.
b.
The boxer threw NP.
c.
The aardvarks throw NP.
d.
Throw NP!
The merge order of arguments therefore depends directly on the semantic composition of
the event.
Marantz points out in a later article (1993) that applied objects pattern with
subjects in regard to this asymmetry rather than with objects. The semantic role of the
theme object does not change when the applied object is varied. In both of the following
!40
examples, the ‘NP’ is cooked by being placed in an oven. The event remains unchanged
in (55b) even though the sentence is pragmatically odd, given that the rock is the
recipient and beneficiary of the baked NP.
(55)
a.
I am baking Mary NP.
b.
I am baking the rock NP.
Marantz concludes the applied object, unlike the theme, is not involved in defining the
event. Instead, it helps identify and distinguish the event in question from other events of
the same type. The addition of an applied object does not change the event class, say
from baking bread to baking something else, it just identifies a specific instance of the
event within that class via Event Identification. For example, if the NP in (55) is bread,
then both cases represent a specific instance of baking bread. Adding the applied object,
Mary or Sally, helps to exclude the other cases of bread baking in the model so that one
can be identified.
It is Marantz’ assertion that the semantic composition of the event and the
affectedness of objects is directly reflected in the syntactic structure of a sentence. Rather
than using a mapping principle like the Uniform Theta Assignment Hypothesis (UTAH)
(Baker 1988), merge order is determined by the sequential affectedness of the object. The
theme is affected within the event and an applied object is affected outside the event. The
applied object must be added after the theme object because it is affected later in the
semantic derivation. Translating the semantic composition of the event to syntax,
elements ‘merging later’ semantically merge higher in the syntactic structure according to
Marantz.5 This means the theme merges before the applied object.
5
Marantz actually assumes the theme object is generated in the specifier position of the minimal VP, not
the complement as shown here. However, the difference between complement and specifier disappears in
Bare Phrase Structure. In the tree, the theme is the daughter of VP, so in some sense it is the specifier.
!41
(56)
Building off the principle of Semantic Compositionality, Marantz proposes the applied
object is introduced by a second functional head or light verb named Appl that functions
to relate the new object to the event.
The example in (55) would therefore have the structure given in (57).
(57)
Marantz’ Semantic Compositionality approach captures the event semantics directly in
the syntax without the need for a mapping principle between a semantic/thematic
representation and the syntactic representation such as UTAH (Baker 1988) and a
stipulatory thematic hierarchy (see Baker 1996, Kirparsky 1987, Machobane 1989,
Jackendoff 1977, Grimshaw 1990, and Larson 1988).
!42
While this approach accounts for most examples of applicatives, Marantz
mentions that instrumental and locative objects in some languages show opposite ccommand relations than benefactive and dative applicatives. Remember from 2.1.3. that
in Kinyarwanda, a quantified theme c-commands an applied instrumental object
(repeated from 50a).
(58)
a.
N-a-fúngul-ish-ije
buri muryangoi úrufunguzo rwáwoi.
I-PST-open-APPL-ASP each door
key
its
‘I opened each doori with itsi key.
(A. Rutayoberana, p.c.; cited in McGinnis 2005:195)
Marantz argues that instrumental and locative objects are affected differently than
benefactive or dative objects. Since both instrumental/locative and theme objects are
affected within the event as opposed to being affected by the event like benefactive
objects, either the applied object or the theme can merge first. He proposes the following
to account for the alternative merge order in instrumental and place locative applicatives.
“Affected object benefactives are compositionally outside the event constructed
by the verb and theme/patient; affected object instruments and place locatives are
affected inside this event and thus may be compositionally inside or outside the
combination of the verb and theme/patient (1993:123-124).”
Allowing the instrument to merge before the theme, as shown in (59), solves the
problems of word order and c-command in instrumental applicatives, but it necessitates
that the theme be selected by Appl, and lie outside the event defined by the minimal VP,
as seen in (59b).
!43
(59) a.
b.
However, merging the theme outside the minimal VP is precisely what Marantz
argues against in 1984 when he showed themes have a much closer relationship to the
verb than subjects and applied objects, such as benefactive objects. Even assuming
instrumental and locative objects are affected by the event differently than benefactive
objects, as Marantz does, the placement of the theme is still problematic. If the theme can
appear in the same position as the VP-external objects like datives and benefactives, then
the asymmetry between applied objects and direct objects explained above would be
mysterious (see example 55).
Marantz is not the only researcher to argue that instrumental and locative objects
are similar to themes (Gruber 1965, Jackendoff 1987, Gropen et al. 1991). Looking
however at how instrumental and locative objects behave in the subject/object dichotomy,
they fall somewhere in between, not fully behaving like subjects and benefactive applied
objects but not like theme objects either. Changing the instrument or place locative
object, for example, does not change the event described or the semantic role of the
theme object, whatever it may be. The ‘NP’ in both (60) and (61) is propelled through the
air regardless of how the action was initiated or where it occurred.
(60)
6 A ball
a.
I threw NP with a ball launcher6.
b.
I threw NP with my hand.
launcher is used to increase the throwing range of a ball, typically when playing with dogs.
!44
(61)
a.
I threw NP in the park.
b.
I threw NP at the stadium.
So instruments and locatives don't look like themes for event composition, but they can
incorporate into adjectival passives, unlike benefactive and dative (goal) objects which
distinguishes them from VP-external objects (examples from Marantz 1993:147).
(62)
a.
hand-made cookies
instrument
b.
spoon-fed children
instrument
c.
home-made cookies
locative
d.
*children-baked cookies
benefactive
e.
*boss-given flowers
goal/benefactive
Assuming the instrument or locative can merge before the theme because it is
sequentially affected before the theme leads one to expect it should act like a theme. In
the same logic, if the theme merges outside the initial event, or the minimal VP in
syntactic terms, then it should behave like a benefactives and datives. Merging
instruments (and locatives) in different positions fails to capture their in-between nature.
Despite this problem, Marantz’ intuitions about themes and applied objects are valuable
and will be considered in greater depth in section 2.4. and again in chapter 4.
2.2.3. Pylkkänen
Pylkkänen (2002, 2008) expands Marantz’ analysis of applicatives. She agrees that
argument structure, and ultimately object position, stems directly from the semantic
composition of the event. Nonetheless, she shows there is a split in the syntactic and
semantic behaviour of applicatives cross-linguistically that a single Appl position is
unable to capture. First, in some languages (e.g. Venda), applicatives are compatible with
unergative verbs, while in other languages, they are only compatible with certain
transitive verbs, like English (examples from Pylkkänen 2008:2).
!45
Venda
(63) Mukasa o-amb-el-a
Mukasa 3SG.PST-speak-APPL-FV
‘Mukasa spoke for Katonga.’
Katonga.
Katonga
English
(64) *Mary spoke Sue.
Intended meaning: ‘Mary spoke for Sue.’
Second, in some applicatives, the applied object is semantically related to the
event while in others, the objects are related to each other. In the Chaga example below,
the applied object mkà ‘wife’ is not related to the theme object kélyá ‘food’. Instead, the
applied object is related to the event of eating food. Contrast this with the English
example, in which Bill is in some way related to a letter, in this case the ultimate
possessor of the letter.
Chaga
(65) N-a-i-lyì-í-à
m-kà
FOC-1SG-PRES-eat-APPL-FV
1-wife
‘He is eating food for his wife.’
k-élyá.
7-food
(Pylkkänen 2008:11)
English
(66) I wrote Bill a letter.
Pylkkänen points out that mkà ‘wife’ cannot enter into a relation with the object kélyá
‘food’. The wife benefits from her husband eating the food but she cannot, for example,
possess the food. In the English example, Bill does not simply benefit from the action of
me writing a letter. I must intend that Bill receives the letter.
Assuming a single position in which the “extra” object is generated, which
Marantz does, does not explain the clear split in applicatives. Pylkkänen proposes that, in
addition to Marantz’ Appl, which she calls High Appl, there is a second type of Appl,
!46
Low Appl. High Appl relates an individual with an event and takes the VP as its
complement and the applied object in its specifier. Being above the VP, High Appl is not
sensitive to the transitivity of predicates and is compatible with both unergative and
transitive verbs. This is the type of applicative seen in the Venda and Chaga examples,
(63) and (65) respectively.
(67) High Appl
On the other hand, Low Appl relates an individual to an individual, encodes a transfer-ofpossession interpretation and is found within the minimal VP below the lexical verb. Low
Appl is not compatible with an unergative predicate because it relates two individuals and
unergative verbs do not have a theme to which the applied object can be related.
Pylkkänen also argues that low applicatives always have a transfer-of-possession of the
theme either to or from the applied object. This means Low Appl will only be compatible
with verbs which allow a possessive interpretation. Pylkkänen cites English applicatives
as an example of Low Appl.
(68) Low Appl
!47
As for the c-command properties seen in applicatives, High and Low Appl
structures both account for the fact that applied objects asymmetrically c-command
theme objects, as seen in many languages. Whether Appl is below the lexical verb or
above it, the applied object is always in the higher position and c-commands the theme
object. Low Appl structure is shown using English (69) and High Appl structure is shown
using Swahili (70).
Low Appl - English (repeated from 37)
(69) a. I sent every workeri hisi paycheck.
b. *I sent itsi owner every paychecki.
c.
High Appl - Swahili (repeated from 48)
(70) a.
Ni-li-m-som-e-a
[kila
SP-PST-OP-read-APPL-FV each
‘I read each author his book.’
b.
mwandishi]i kitabu
author
book
chakei.
his
*Ni-li-m-som-e-a
[kila kitabu]i mwandishi wakei.
SP-PST-OP-read-APPL-FV each book
author
its
‘I read for its author each book.’
(Marantz 1993:117)
!48
c.
The High/Low Appl Theory accounts for the typological split in applicatives, whether
they are compatible with unergative predicates or not, and the word order and ccommand properties exhibited by benefactive and dative applicatives.
Although insightful, High and Low Appl are not able to account for the full range of
applicatives. Instrumental applicatives, noted earlier (see section 2.1.3), show the theme
c-commands and precedes the instrument which is the opposite pattern of what High and
Low Appl predict. Recall that Marantz proposes that the theme and the applied
instrument may merge in either order. This is not an assumption Pylkkänen makes. An
applied object must be generated in the specifier of the Appl phrase, whether Low or
High.
Given instrumental and locative applicatives are compatible with intransitive
predicates, she assumes they are high applicatives.
Kindendeule Instrumental
(71) a-ki-tyang-i
hi-latu.
1-PST-walk-APPL 8-shoe
‘He walked with shoes.’
(Ngonyani 1998:72)
Kindendeule Locative
(72) Yesu a-ki-hwel-e
ku-Gɔlgɔta.
Jesus 1-PST-die-APPL 15-Golgota
‘Jesus died at Golgota.’
(Ngonyani 1998:73)
!49
The problem with this approach is that in a High Appl structure, the instrumental object is
predicted to precede and c-command the theme object, contrary to what is seen. The
instrument ícyúuma ‘knife’ follows the theme inyama ‘meat’ in (73).
Instrumental applicative - Kinyarwanda
(73)
Umubooyi
a-ra-kat-iish-a
inyama
ícyúuma.
cook
she-PRES-cut-APPL-ASP
meat
knife
‘The cook is cutting meat with a knife.’
(Kimenyi 1980:32)
A movement analysis, such as rightward movement of the instrumental object in
(73) cannot save the High Appl analysis. In benefactive applicatives, the benefactive
object can appear to the right of the theme. Marantz argues that in Kinyarwanda
benefactive and dative applicatives, applied objects can move to the right via rightward
adjunction. This allows the applied object to appear to the right but still in a position ccommanding the theme, even from the right.
(74)
Ni-li-m-som-e-a
kitabu
SP-PST-OP-read-APPL-FV
book
“I read each author his book.”
chakei
his
[kila
each
mwandishi]i.
author
So rightward movement of the instrument in instrumental applicatives is certainly
plausible and would explain the word order facts. However, assuming a rightward shift of
the instrument, via the same process as the benefactive movement in (74), fails to explain
the c-command relationship between the instrument and the theme. Under such an
analysis, the instrument is still expected to c-command the theme from the right, which it
does not. Using quantifier binding, (75) shows the instrument, buri rufunguzo ‘each key’
cannot c-command the theme, umuryano wáyo ‘its door’.
!50
Kinyarwanda (sentence repeated from 49b)
(75) b.
N-a-fúngul-ish-ije
umuryano
wáyo*i/j
I-PST-open-APPL-ASP door
its
‘I opened itsi door with each keyj/*i.’
buri
rufunguzoi.
each
key
Additionally, the obligatory nature of such a shift remains mysterious. Nothing in the
High/Low Appl Theory explains why instruments would have to appear to the right while
benefactive and dative objects can stay to the left or optionally move to the right.
In sum, neither a High nor Low Appl analysis accounts for the properties of
instrumental and locative applicatives. I argue in chapter 4 that they involve a third, new
type of applicative head. The problem of instrumental and locative applicatives is picked
up by McGinnis (2005).
2.2.4. McGinnis
McGinnis (2005) builds off of Pylkkänen’s (2002) High and Low Appls, assuming they
are responsible for applicative formation, but arguing that High Appl can merge either up
or down.7 She shows that problematic instrumental and locative applicatives in some
Bantu languages can be accounted for using High Appl if it merges downward with the
VP.
McGinnis proposes that derivations proceed outward, starting with the verb.
Elements can merge up or down depending on their lexical specification. She argues that
allowing syntactic items to merge up or down does not violate UTAH. For McGinnis,
UTAH is not a constraint on representations but a condition on External Merge.
7
Downward Merge is based on work by Philips (2003). Philips argues that syntactic derivations proceed
top-down, based on data from constituency tests. While intriguing, a top-down approach to syntax, in
which successive elements merge downward from the previously established node, is problematic because
it does not match with generative theories of semantic derivation, including Semantic Compositionality and
UTAH which are based on a bottom up derivation. Major works on the semantic nature of argument
structure (Marantz 1993, Kratzer 1996, and Pylkkänen 2008) assume a bottom-up derivation beginning
with the predicate.
!51
McGinnis (2005) notes the recent movement away from predefined phrase
structures in Minimalist theories. Thematic relations between items are defined at Merge
depending on the nodes involved and not sisterhood relations in the completed structure.
Under McGinnis’ conception of UTAH, the thematic role of an object doesn’t depend on
its final syntactic position in a structural representation but on the node with which it
merges. Two elements can merge without necessarily becoming sisters. The figure in
(76a) shows the representation of Appl that has merged upward with the VP. The figure in
(76b) shows the representation of Appl that has merged downward with the VP.
(76)
a.
b.
The NPtheme receives the theme role because it merges with V in both cases. This
relationship is obvious in (76a) but more opaque in (76b) since the subsequent merge of
the Appl head leads to a reanalysis of the structure; the NPtheme becomes the specifier of
Appl. To illustrate this, the first step of the derivation of (76b) is given in (77a). The NP
merges with the verb and is assigned the theme theta role. In the next step, illustrated in
(77b), Appl merges with the VP. However, since the VP already has a complement,
something has to move in order to allow Appl to merge down. Appl takes the place of the
NPtheme and subsumes it in its specifier. In the last step, shown in (76b), the applied object
has merged downward with Appl placing it below the theme object. The original
relationship created by the merge of V + NPtheme is not broken by the reanalysis and so
the NP remains a theme even though it is no longer the sister of V.
!52
(77) a.
b.
Since Appl merges outside the VP, its object is generated outside of the event as shown in
(76b). Although the final structure in (76b) looks like the structure of Low Appl, the
downward merging High Appl does not relate two individuals like Low Appl does. It
relates an individual to an event by merging with the VP and introducing an object.
Moreover, unlike Low Appl, this Appl is insensitive to the transitivity of the predicate.
Dative applicatives are not the focus of the article, but McGinnis does assume
they are formed using Low Appl. As will be shown in the next section, this assumption is
problematic because Low Appl is unable to account for floating quantifiers in dative
applicatives.
2.2.5. Georgala
Another proposal that builds off of Marantz’ Appl Theory and Pylkkänen’s Theory of
High/Low Appl, but in a different direction than McGinnis, is the Thematic/Raising
Applicative Hypothesis by Georgala (2012). She argues that Low Appl cannot account
for several particular properties of dative applicatives. She adopts High Appl as proposed
by Pylkkänen but argues that Low Appl be rejected. The main motivation for rejecting
Low Appl is that it fails to account for floating quantifiers in dative applicatives.8 In the
following example from Mandarin, the dative object háizimen ‘children’ precedes the
quantifier měi-rén ‘every’ even though the quantifier scopes over the dative object.
8
Georgala also uses the false entailments that Low Appl leads to, as pointed out by Larson (2010) and
morphology to motivate rejection of Low Appl. I will discuss Larson’s criticism in 3.4 in greater detail.
Data from floating quantifiers suffice for the current discussion.
!53
Chinese - Floating quantifier
(78)
Wǒ
sòng-gěi
háizimen [měi-rén]
1SG
give-GEI
children
[yībǎi
every(one) 100
kuài
qián].
CL1
dollar
‘I gave the children each 100 dollars.’
(Paul and Whitman 2010:279)
Georgala also cites an example from German in which the dative object appears to
the left of an adverb merged at the VP level. Assuming the adverb heimlich ‘secretly’ is at
the left edge of VP, then it provides evidence that the dative object den Studenten ‘the
students’ is outside of the VP.
(79)
Der
the.NOM
Hiwi
teaching.assistant
hat
has
den
the.DAT
Studenten
VP[heimlich
students.DAT secretly
VP[einen
alten
Test
ausgeteilt]].
an.ACC
old.ACC
quiz.ACC distributed
‘The teaching assistant secretly distributed an old quiz to the students.’
(Georgala 2012:74)
The floating quantifier in (78) and the adverb placement in (79) are difficult to explain
assuming a Low Appl structure. How does the dative object end up outside the minimal
VP?
Instead of Low Appl and High Appl, she proposes two Appls that merge in the
same position, as the sister to VP. The result is a single applicative structure reminiscent
of Marantz (1993) but two heads differing for selectional properties, which explains the
division of applicatives into two groups, reminiscent of Pylkkänen’s approach (2008).
Georgala calls the two types of applicatives Thematic and Raising. The head in thematic
applicatives is called Thematic Appl (ApplT) and the head in raising applicatives is called
Expletive Appl (ApplE). ApplT is equivalent to High Appl; it relates an individual to an
event. It does this by selecting the applied object and linking it with the VP in its
complement. Raising applicatives involve an expletive head, ApplE and has no semantic
!54
content, unlike ApplT. Rather, in raising applicatives, the dative object is selected by the
verb. It is the verb that relates the dative object to the theme object. The dative object
then moves to the ApplE projection for licensing.
(80) a. Thematic Applicative
b. Raising Applicative
This means that in raising applicatives, the object is not truly applied to the
predicate in the same sense as thematic applicatives. Nonetheless, datives are still
considered applicative structures because they contain an Appl head, ApplE. Dative
applicatives, including goals, recipients, and sources, are raising applicatives and the
other types of applicatives are thematic according to Georgala. Two examples from Greek
are provided below.
Greek Dative - Raising applicative
(81) O
Nikos
edhose
tis
Marias
ena
the.NOM Nick.NOM gave.3SG the.GEN Mary.GEN a.ACC
vivlio.
book.ACC
‘Nick gave Mary a book.’
Greek Benefactive - Thematic applicative
(82) O
Nikos
fitepse
tis
the.NOM Nick.NOM planted.3SG the.GEN
Marias
luludhia
…
Mary.GEN flowers.ACC
‘Nick planted flowers for Mary…(in the garden.)’
(Georgala 2012:104)
!55
While these two sentences appear to be similar on the surface, the motivation for
adopting two derivations instead of one comes from semantic differences and the position
of floating quantifiers. Greek dative applicatives are argued not to be thematic like
benefactive applicatives, but rather raising applicatives because they show the properties
of low applicatives such as the transfer of possession between the objects and are only
compatible with transitive predicates (Georgala 2012). In raising applicatives, the objects
have an underspecified relationship resulting from their positions in the complement and
specifier of V. The transfer of possession interpretation itself comes from the lexical
semantics of the verb. As a result, raising applicatives are only possible with verbs that
are compatible with or encode a transfer of possession.
Benefactives, on the other hand, do not show the properties of low applicatives
and are not limited to verbs with transfer of possession semantics. They are compatible
with intransitive and stative predicates as well as transitive predicates indicating they are
not selected by the verb.
Stative predicate + benefactive applicative
(83) Borite na kratisete tis
Marias
afto
to
forema…
can.2PL to keep.2PL the.GEN Mary.GEN this.ACC the.ACC dress.ACC
‘Can you keep this dress for Mary…(until tomorrow?)’ (Georgala 2012:106)
The compatibility with intransitive and stative predicates motivates a thematic applicative
analysis. Raising applicatives are not compatible with these predicates.
Returning now to examples of floating quantifiers in dative applicatives, this
phenomenon is easily accounted for by a raising applicative analysis. In the following
example from Greek, the dative object, tus pelates ‘the customers’ is modified by a
floating quantifier, olus ‘all’.
!56
Floating quantifier - Dative applicative
(84) Servira
tus
pelates
olus
served.1SG the.ACC customers.ACC all.ACC
I served all customers breakfast.’
proino.
breakfast.ACC
(Georgala 2012:127)
Georgala assumes a Sportiche style analysis of floating quantifiers where the quantifier is
merged together with the dative object in the specifier of VP (Sportiche 1988). The dative
object then moves to the specifier of ApplEP, leaving the floating quantifier in the
specifier position of VP. If the dative never moves from the VP, as Pylkkänen assumes,
the word order in (84) would be hard to explain. The tree in (85) illustrates the relevant
structure.9
(85)
The raising applicative structure successfully accounts for both the semantic transfer of
possession from the lexical semantics of the verb and the movement of the dative object.
As with previous analyses, instrumental and locative applicatives remain problematic for
the Thematic/Raising Hypothesis. Both applicative types predict that the applied object
will precede and c-command the theme object, as is easily seen in the structures in (80).
9
The final form of the verb is given in the structure in (85) even though the inflectional domain has been
truncated. This is simply for easy comparison with the sentence in (84).
!57
Thematic Appl falls victim to the same problems as High Appl, not surprisingly, since
they are structurally the same. The instrumental object cannot be added such that the
theme c-commands it and thus the structure fails to explain instrumental applicatives in
languages like Kinyarwanda. Raising applicatives have the same problem as the position
of Appl is the same so it cannot account for the theme c-commanding the instrument
either. Raising Appl encounters another problem with instrumental and locative
applicatives: it fails to account for the compatibility of instrumental and locative
applicatives with intransitive predicates.
2.2.6. Summary
The approaches presented in this section have much to offer in the way of analyzing
applicative sentences. However, certain applicatives resist explanation for each analysis.
Marantz’s (1993) approach captures the problematic instrumental data but raises
theoretical issues like allowing the theme to be generated in the same position as the
applied object. If the theme can be generated outside the VP, then a syntactic explanation
of the difference between core objects and external objects disappears. Identifying the
need for more than one applicative structure is a major step forward, but Pylkkänen’s
High and Low Appl structures (2008) fail to account for the instrumental data from
Kinyarwanda and other similar data. Additionally, the definition of Low Appl is also
problematic (Larson 2010). The final problem with the High/Low Applicative analysis is
that neither Low Appl not High Appl are able to capture the phenomenon of floating
quantifiers in dative applicatives (Georgala 2012). McGinnis (2005) extends Pylkkänen’s
analysis by arguing that High Appl can merge downward with the VP. This modification
does capture the word order and c-command effects seen in Kinyarwanda instrumental
applicatives but is not able to account for dative applicatives. While not the focus of the
article, McGinnis does assume Low Appl is used in dative applicatives, which is
problematic as Georgala (2012) and Larson (2010) point out. Georgala successfully
accounts for dative applicatives and floating quantifiers with the Raising Applicative
!58
analysis. However, again, instrumental applicatives remain problematic. Before outlining
my proposal for Wolof applicatives, I turn to previous work done on Wolof applicatives.
2.3. Existing work on Wolof applicatives
Many works on Wolof focus on clause structure, as mentioned in Chapter 1. Here I
present previous works that touch on applicatives in Wolof. Nouguier-Voisin (2002) and
Creissel and Nouguier-Voisin (2004) focus on valency changing suffixes including
applicatives. While they provide ample data and description of Wolof applicatives, they
focus their analyses on the historical development of these sentences and their
morphology. Schwartz (1975) does not focus on applicatives in his article, but he does
present many applicative sentences and discusses different object properties. Dungian’s
doctoral thesis (1994), which examines the clause structure of Wolof, describes and
analyzes applicatives in chapter 6. Dione (2013) treats Wolof causative and applicative
polysemy in the Lexical Function Grammar framework. I focus on formal analyses of
Wolof applicatives so I discuss only Schwartz, Dunigan, and Dione’s works below.
2.3.1. Schwartz 1975
Schwartz looks at the properties of objects in Wolof and argues based on certain tests that
Wolof does not have a direct object. The tests he uses are properties commonly associated
with the direct object of a transitive verb like post-verbal position, case-marking,
passivization, object incorporation, etc. What he finds is that these “direct object
properties” can be associated with either the DO (what I call the theme) or the IO (what I
call the applied object) in double object constructions. Given that there is not one object
alone to which direct object properties can be ascribed, he argues there is no direct object
in Wolof. Though this assertion has not been supported by later works on Wolof, the data
he provides is useful for exploring applicatives. One drawback to this article is that the
!59
dialect of Wolof is not specified. As will be seen in chapters 3 and 4, the speakers
consulted for this thesis often had different intuitions about Schwartz’ sentences as to
their meaning and grammaticality.
Schwartz shows that DO and IO object in Wolof show symmetrical behaviour in
relation to their properties. He ascribes both objects with the following properties:
position of applied object (verbal adjacency), pronominalization of the objects,
reflexivization, clefting, and topicalization (tests not relatable to applicatives are not
included). The majority of data come from dative applicatives. Instrumental, locative, and
benefactive applicatives are assumed by Schwartz to behave the same.
Table 1. Object properties exhibited by DO and IO in Wolof
position
pronominalization
reflexivization
clefting
topicaliztion
DO
✓
✓
✓
✓
✓
IO
✓
✓
✓
✓
✓
The property “position” refers to the ability of the object to appear adjacent to the verb.
In Wolof, either the DO or the IO can immediately follow the verb. Pronominalization
means that both the DO and IO are replaced with the same form of the object and
interrogative pronouns. As for reflexivization, he refers to the ability of the DO and IO to
be a reflexive pronoun. As for clefting and topicalization, both the DO and IO can be
fronted. While Schwartz interprets this to mean that there is no single DO in Wolof since
the properties can be ascribed to NPs with different theta roles (recipient, benefactive,
instrument, or locative), we now know this behaviour is precisely what is seen in
symmetrical applicatives. As will be illustrated in chapter 4, however, not all applicatives
in Wolof are symmetrical.
2.3.2. Dunigan 1994
In chapter 6 of her dissertation, Dunigan (1994) describes four types of Wolof
applicatives: dative, benefactive, instrumental, and locative. She also notes some of these
!60
constructions are inherently ditransitive, requiring no applicative morphology (86 and
87), while others are lexically transitive and require an applicative suffix to be
grammatical (88-89).
Dative applicative
(86) Faatu jox-na
Fatou give-AFF.3SG
yaay-am
mother-POSS.3SG
‘Fatou gave her mother the kola nut.’
Locative applicative
(87) Faatu gunge-na
Fatou accompany-AFF.3SG
xale
child
guru
kola.nut
gi.
DEF1
(Dunigan 1994:237)
bi
seen
DEF1 POSS.3PL
‘Fatou accompanied the child to their house.’
ker.
house
(Dunigan 1994:238)
The number of verbs that are inherently ditransitive in Wolof is limited. Only dative
applicatives and at least one locative applicative allow two objects with no additional
morphology on the verb. The majority of applicative sentences in Wolof, however, do
require an applicative morpheme. The suffix that allows benefactive objects is -al. In
benefactive applicatives, the benefactive precedes the patient.10
(88)
Faatu
Fatou
sampa-al-na
yaay-am
(>sampalna)
build-APPL-AFF.3SG mother-POSS.3SG
ker.
house
‘Fatou built her mother a house.’
(Dunigan 1994:238)
The applicative suffix for instrumental and most locative applicatives has the form -é.11
Dunigan notes that the order of the objects differs from the benefactive example. In
instrumental applicatives, the patient precedes the instrument.
10
Dunigan refers to the original direct object as the ‘patient’. I use the term ‘theme’ in my analysis but for
now, the two terms are treated as equivalent.
11
Dunigan spells the suffix with the accent -é. Some orthographies spell it without the accent. The latter
spelling will be adopted in chapters 3 and 4.
!61
(89)
Xale bi
bindd-é
në
letar bi
estilo.
child DEF1 write-APPL AFF.3SG letter DEF1 pen
‘The child wrote the letter with a pen.’
(Schwartz 1975:226, cited from Dunigan 1994:239)
Dunigan cites the instrumental example, given by Schwartz (1975) in his article
on direct objects in Wolof, but notes that her consultant, a speaker of Banjul Wolof,
rejected the sentence in (89). Her consultant only accepted sentences in which the
instrument was contained in a prepositional phrase. The presence of the applicative suffix
-é in conjunction with the preposition is optional (90).
(90)
Gaan(-é)
wound(-APPL)
na
AFF.3SG
Modu
Modu
‘He/she wounded Modu with a knife.’
ak
with
paaka.
knife
(Dunigan 1994:239)
The use of the instrumental -e suffix will be discussed in more detail in chapter 4. As for
morphologically derived locative applicatives, Dunigan does not cite any examples but does
mention they exist.
After providing a description of basic applicative sentences, Dunigan compares
the object properties of each type. Her findings are summarized in the following chart.
Dunigan presents data that all four types allow alternate word order for the objects.12
However, she does conclude that default word order for benefactive and dative
applicatives is applied object - theme object and the default order for instrumental and
locatives is theme object - applied object. Benefactive and dative applicatives allow both
objects to be replaced by an object marker, wh-moved, or a combination of those two
properties. Instrumental and locative applicatives are only shown with the instrumental
and locative objects undergoing object marking and wh-movement so it is unknown if the
theme object also shares these properties.
12
It will be shown in chapter 4 that this is not true of instrumental and locative applicatives for Saint Louis
speakers which is the focus of this dissertation.
!62
Table 2: Object properties as described by Dunigan 1994
Benefactive
Dative
Instrumental
Locative
Preferred word
order
BEN > PAT
REC > PAT
PAT > INST
PAT > LOC
Alternate word
order
✓
✓
✓
✓
Object marker
BOTH
BOTH
INSTR*
LOC*
Wh-movement
BOTH
BOTH
INSTR*
LOC*
Object marker with BOTH
wh-movement
BOTH
NO DATA
NO DATA
*No examples in which the patient was object marked or wh-moved were given.
The analysis Dunigan adopts for Wolof builds off of Larson (1988) and Aoun and
Li (1989)’s work on applicatives in English. She assumes a complex predicate structure
involving two verbal phrases as Larson proposed (1988) but she assumes the objects are
base generated in their surface structure positions following Aoun and Li (1989). She
adopts Marantz’ Semantic Compositionality and assumes the applied object merges
higher than the theme but does not assume the applied object is introduced by the second
verbal head. Instead, she generates it within the minimal VP.
(91)
!63
As for cases of non-default word order in benefactive and dative applicatives, she
assumes the higher (applied) object is demoted to an V’-adjunct position to the right of
the verb. This is similar to the rightward adjunction proposed for Bantu languages.
The structure for instrumental and locative applicatives is similar to the one seen
above except that the applied object in instrumental and locative applicatives is generated
as a V’-adjunct. The theme moves to the specifier of VP and the lower V’ gets
reanalyzed13 as a V and assigns case to the applied object.
(92)
This second structure is proposed for instrumental and locatives in order to account for
the word order of instrumental and locative applicatives, which is inverted from
benefactive and dative applicatives. As for cases of the non-default word order, they
results from the locative or instrumental object moving to the specifier of VP while the
theme stays in the V complement position. Dunigan does not assume, as Marantz does,
that an applied instrumental or locative object can merge as the sister of V. The theme
always merges in that position.
The symmetrical properties of all types of applicatives results from both objects
being assigned their theta role and case by the lexical verb. Since both get case from the
13
For more on V’ reanalysis in applicatives see Larson 1988.
!64
lexical verb, Dunigan assumes they both receive the same case, accusative. It is the
accusative case that allows them to be object marked and wh-moved.
2.3.3. Dione 2013
Dione describes and analyzes applicative and causative sentences in Wolof from the
perspective of Lexical Functional Grammar (LFG). His goal is to provide an analysis of
the effect of these affixes on argument structure and to highlight the differences between
causatives and applicatives despite the identical morphology.
Dione shows that beneficiary and recipient applicatives are obligatory, meaning
objects with these roles are expressed only as an applied object in applicative sentences
and cannot appear in prepositional phrases. The other remaining types are optional
applicatives because they can be expressed using a prepositional phrase or as an applied
object. As for object properties, he argues benefactives must be adjacent to the verb and
are the non-restricted object. Instrumental applied objects must follow the theme and are
restricted. We will see in chapter 3, that the data I present don’t completely match his.
He analyzes applicatives as complex predicates with the following (LFG) astructure.
(93)
Applicatives a-structure:
‘PRED1<%PRED, ARG>’
ARG: any semantic role
introduced by the applicative
The notation %PRED stands for “a variable to be filled in by a predicate’s a-structure of
the non-derived verb (Dione 2013)”. Within this structure, he proposes four distinct types
of a-structures for applicatives.
His goals are somewhat different than mine, given the framework. He is interested
in argument structure while I am interested in syntactic structure, in particular, the ccommand relationship between the theme object and the applied object. As such, it makes
!65
comparing approaches very difficult. However, I do agree with him that different
analyses are required for the different types of applicatives.
In the following chapters, I adopt a generative approach to applicatives following
Dunigan, Marantz, Pylkkänen, McGinnis and Georgala. We will see that a more finegrained approach to applicatives is needed. I now turn to a sketch of the analyses that will
be proposed in chapters 3 and 4.
2.4. Theoretical Assumptions
In what follows, I adopt the Minimalist Program within Generative Grammar (Chomsky
1993, 1995, 2000). I assume Bare Phrase Structure, which means I assume vacant
positions are not projected, contra X-bar theory. I also reject a syntax that is strictly built
from the bottom up (Kratzer 1996, Chomsky 1995, 2000), adopting instead a syntax that
is built from the verb outward (McGinnis 2005). This means that new items can merge
either up or down. I follow the Minimalist Program in assuming Merge can only take
place at the highest node. Put differently, Merge takes place at the most recently
generated node. The distinction in terminology between ‘highest node’ or ‘most recently
generated node’ will be particularly relevant for the discussion of oblique applicatives in
chapter 4.
I assume that derivational heads, like Appl, are defined just like lexical items for the
arguments they select and the direction in which they merge and features they carry. I
assume a feature driven syntax, which means syntactic operations like movement are
motivated by features such as EPP feature, wh-feature, and so on. Agree is assumed to be
established between a probe, P and a goal, G within the domain of the probe, D. Case is
‘assigned’ to NPs within their domain via Agree. I assume that Merge (External Merge)
must take place before Move (Internal Merge) (Chomsky 1995, 2000). This ordering will
become relevant when discussing the object properties of applicatives in chapters 3 and 4.
!66
2.4.1. Overview of proposed analysis
In this thesis, I extend the proposal that there is more than one applicative head
(Pylkkänen 2008, Georgala 2012). I argue for three types of applicative heads rather than
two. First, I argue that benefactive applied objects are selected by Thematic Appl.
Second, dative applied objects, which include goals, recipients, and sources, are selected
by the lexical verb itself and involve movement of the dative object to an expletive
applicative projection, ApplE. This is what Georgala calls a Raising Applicative. ApplE is
different from ApplT in that it has no semantic content and does not not select an object,
while ApplT does. Finally, I propose that oblique applied objects, such as instruments and
locatives, involve a third, new applicative head which I call Oblique Appl (ApplO).
ApplO is different from ApplT because it merges down from the VP level placing it lower
than the verb and any theme object present, following the work of McGinnis (2005).
ApplO is different from ApplE in that it has semantic content, selects an argument, and
assigns it a theta role.
The three structures proposed are as follows:
(94) Thematic Applicative
I adopt Georgala’s thematic applicative for benefactive applicatives in Wolof. The theme
merges with the verb. ApplT merges and selects the benefactive object. The benefactive
object receives Case from v and the theme receives Case from ApplT.
!67
Dative applicatives in Wolof are of the raising applicative type. The verb selects
both the theme and dative objects. The dative always merges above the theme since it is
affected by the event defined by the V+Theme combination. ApplE merges with the VP
and the dative object raises to ApplEP for licensing. The applied object is licensed and
receives Case from v while the theme object receives case from ApplE.
(95) Raising Applicative
The structure of raising applicatives, illustrated in (95) is similar to the structure proposed
for Wolof benefactive and dative applicatives by Dunigan, illustrated in (91). In both
structures, the applied object is generated as the specifier of the verb. The difference is
that Dunigan assumes that both benefactive and dative objects are generated in
[Spec,VP], while I follow Georgala in assuming only dative objects are. I argue that
benefactive applied objects are thematic applicatives. Additionally, Dunigan does not
assume movement of the dative object to the higher Appl projection.
The derivation of ApplO involves downward Merge and follows McGinnis
(2005). The lexical verb merges first with the theme object. Then ApplO merges with the
VP node in a downward fashion and the theme is reanalyzed as part of the ApplO phrase.
The instrumental or locative applied object then merges, again downward, placing it in
the domain of the theme. The instrumental or locative applied object receives Case from
ApplO since it are within its domain and the theme receives Case from v.
!68
(96) Oblique Applicative
Dunigan’s proposal for instrumental and locative applicatives is similar. In both
structures the theme merges with the verb first and then the instrumental or locative
merges and the instrument or locative is in the lower object position by the end of the
derivation. Dunigan accomplishes this by raising the of the theme while I assume, as in
(96), the downward merge and reanalysis places the instrument lower than the theme.
I will show in chapter 4 how the ApplO analysis explains the in-between nature of
instruments and locatives discussed in section 2.2.2. As oblique applicatives, they merge
outside the VP so they behave similar to applied benefactives and subjects. Since they
merge down they end up within the minimal VP, so they behave similar to themes as
well.
Chapters 3 and 4 contain the detailed analyses and discussion of the proposed
ApplT, ApplE, and ApplO structures.
2.5. Conclusion
In this chapter, I have presented a typology of applicatives from different languages while
focusing on their object and c-command properties. Applicatives, rather than languages,
are classified as symmetrical or asymmetrical based on evidence from Kinyarwanda (and
other languages) where benefactive applicatives are symmetrical but locative applicatives
are asymmetrical. Several major approaches to applicatives within the Minimalist
!69
framework were presented and although promising, they were unable to capture all types
of applicatives cross-linguistically. I then looked at three previous articles on Wolof
applicatives. Finally, I presented a brief overview of the proposed analysis, which will be
discussed in detail in the next two chapters. Chapter 3 focuses on benefactive and dative
applicatives. Chapter 4 treats instrumental and locative applicatives.
!70
CHAPTER 3
3. BENEFACTIVE AND DATIVE APPLICATIVES
In this chapter, I explore the syntactic properties of Wolof benefactive and dative
applicatives in more detail using objecthood tests, quantifier binding, reflexivization, and
weak cross-over tests. Instrumental and locative applicatives will be addressed in chapter
4.
Syntactic tests will show that benefactive and dative applicatives pattern together
for quantifier binding, reflexivization, and weak cross-over but differ for morphology,
selectional restrictions, and semantic interpretation. To account for the differences, I
adopt the distinction between Thematic and Raising Applicatives of Georgala (2012).
More specifically, I argue that Wolof benefactive and dative applicatives involve different
derivations. Benefactive applicatives involve Thematic Appl (ApplT) while dative
applicatives involve the expletive head, ApplE, of raising applicatives.
The chapter is structured as follows: Object properties of benefactive and dative
applicatives are described and compared in section 3.1. Syntactic tests for determining ccommand are evaluated in section 3.2 and shown to be amenable to Wolof. The tests are
then applied to Wolof benefactive and dative applicatives to determine the c-command
relationship between the objects. The analysis of these applicatives is presented in section
3.3.
3.1. Object Properties
As mentioned in chapter 2, the object properties shown in applicative constructions are
useful to classify and describe applicatives. In what follows, I briefly show benefactive
and dative applicatives in Wolof exhibit symmetrical object properties such as free word
order of the objects, pronominalization, and extraction. Passivization data are not
included because Wolof has no syntactic passive construction as we saw in chapter 1.
!71
3.1.1. Word Order
The word order of the objects in benefactive and dative applicatives is variable. I assume
this is an instance of object scrambling as it is a case of a change in word order without
any (apparent) effects on meaning. Either the theme object or the applied object can be
the first object in benefactive and dative applicatives. The benefactive object in (1),
xaritam ‘his/her friend’ (underlined) can either precede the theme object xar mi ‘the
sheep’ or follow it. The meaning does not change between the different orders. The same
thing is seen with the dative object xale bi ‘the child’ in (2).
(1)
(2)
a.
Rey-al
na
xarit-am
xar
pay-BEN 3SG friend-3SG.POSS sheep
‘He/she killed the sheep for his/her friend.’
mi.
DEF1
b.
Rey-al
na
xar
mi xarit-am.
kill-BEN 3SG sheep DEF1 friend-3SG.POSS
‘He/she killed the sheep for his/her friend.’
a.
Damay
jox xale bi
neexal.
1SG.PFOC
give child DEF1 gift
‘I gave the child a gift.’
b.
Damay
jox
1SG.PFOC
give
‘I gave the child a gift.’
neexal xale bi.
gift
child DEF1
These data contrast with the English applicative examples discussed in Chapter 2.
In English, the dative object must precede the theme. Object scrambling is not permitted.
Remember from chapter 2 that English applicatives are asymmetrical.
(3)
a.
Sally gave Bill a book.
b.
*Sally gave a book Bill.
!72
In Wolof, word order between the two objects is free, suggesting the applicatives are
symmetrical.
3.1.2. Pronominalization
In Wolof, there is no ‘object marking’ via verbal prefixes like the languages discussed in
chapter 2. Instead Wolof uses clitics to pronominalize objects and these can be used to
distinguish between oblique and direct objects. The clitic object pronouns are used for
direct objects of the verb. Objects that are contained in a prepositional phrase or in a cleft
or other position not directly following the verb are pronominalized using strong
pronouns. Although direct object clitics do not affix morphologically to the verb when
pronominalized, as in Bantu languages, we can use the distinction between clitic and
strong pronouns to identify oblique and direct objects in Wolof and determine if the
objects have symmetrical or asymmetrical properties.
The use of clitic pronouns in Wolof is similar to the strong and weak pronouns in
French. In French, weak, or clitic object pronouns, like te ‘you’ are used when
pronominalizing arguments of the verb and not complements of prepositional phrases.
They must appear directly to the left of the verb as seen in (4a). If they appear separated
from the verb, in a prepositional phrase, clitic pronouns are not licit (4b). Instead, the
strong pronoun form, toi ‘you’, is required as shown in (4c).
(4)
a.
Je te vois souvent.
I see you often. (lit. I you see often.)
b.
*Je pense souvent à te.
I often think of you.
c.
Je pense souvent à toi.
I often think of you.
!73
This is also true for Wolof objects. Lexical NP objects that appear in a cleft position or
within a prepositional phrase can only be replaced by a strong pronoun while direct
objects are replaced by a clitic pronoun.
The sentences in (5) show a transitive sentence with the direct object jën yi ‘the fish’. The
direct object is pronominalized using the clitic pronoun leen ‘them’ (5b). In contrast, the
oblique object in (6b) must be pronominalized with the strong pronoun moom ‘him/her/
it’. The last sentence (6c) shows the prepositional benefactive object cannot be replaced
by a clitic. Note also, that clitics in Wolof must follow the subject marker, naa in these
examples.
(5)
(6)
a.
Japp
naa
jën
catch 1SG fish
‘I caught the fish.’
wi.
DEF1
b.
Japp naa ko.
catch 1SG 3SG.OBJ
‘I caught it.’
a.
Japp naa jën
yi
ngir
catch 1SG fish DEF.PL for
‘I caught the fish for Ousmane.’
b.
Japp naa jën yi
catch 1SG fish DEF.PL
‘I caught them for him.’
c.
*Japp naa ko
jën
catch 1SG 3SG.OBJ
fish
‘I caught the fish for him.’
Ousmane.
Ousmane
ngir
moom.
PREP1 3SG.OBJ
yi.
DEF.PL
!74
The forms of the strong and clitic object pronouns are given in Table 1.
Table 1: Object Pronouns
Clitic Strong
1SG
ma
man
2SG
la
yow
3SG
ko
moom
1PL
nu
nun
2PL
leen
yeen
3PL
leen
ñoom
Comparing transitive and prepositional complement sentences in Wolof in (5) and (6)
respectively, we see that oblique objects of the verb are pronominalized with strong
pronouns, unlike the direct object.
Turning to pronominalization of objects in applicatives, we see that applied
objects behave the same as the direct object in transitive sentences. Example (7b) shows
the benefactive applied object (underlined) replaced with a clitic pronoun. In (7c) the
theme object toggukay bi ‘the chair’ has been replaced. Note that the same object clitic ko
has been used for both the benefactive and the theme in these examples.
(7)
a.
b.
c.
Daaj-al
naa Badara toggukay
construct-APPL 1SG Badara chair
I constructed the chair for Badara.
bi.
Daaj-al
naa ko
toggukay
construct-APPL 1SG 3SG.OBJ chair
I constructed the chair for him/her.
bi.
DEF1
DEF1
Daaj-al
naa
ko
Badara.
construct-APPL 1SG
3SG.OBJ Badara
I constructed it for Badara.
The same pattern is seen in the dative example in (8b), where the dative applied object
has been pronominalized with the clitic pronoun ko ‘him/her/it’. In (8c) the theme object
!75
has also been replaced by ko. Both objects can be pronominalized with clitics
simultaneously as well (8d).1
(8)
a.
Jox naa xale
bi
neexal.
give 1SG 3SG.OBJ DEF1 gift
I gave a present to him/her.
b.
Jox naa ko
neexal.
give 1SG 3SG.OBJ gift
I gave a present to him/her.
c.
Jox naa ko
give 1SG 3SG.OBJ
I gave it to the child.
d.
xale
child
bi.
DEF1
Jox naa ko
ko.
give 1SG 3SG.OBJ 3SG.OBJ
I gave it to him/her.
The fact that both object can simultaneously be pronominalized will be relevant in
chapter 4. These data illustrate the benefactive and the theme object have the same object
status when it comes to pronominalization, indicating symmetrical applicatives.
3.1.3. Extraction
In this section, I consider data from extraction in applicatives. Before looking at
applicatives, it is important to note that the direct object of a transitive verb can be
extracted for focus (9) and questioning (10) in Wolof.
1
The acceptability of both objects being pronominalized in benefactives (10) is not entirely clear. Some
speakers did not accept the sentence in (i) but this judgement was not unanimous.
(i)
Daaj-al naa ko ko.
The reason for the difference in acceptability between (i) and (14d) is also unclear.
!76
(9)
Jën wi
laa
taxañ.
fish DEF 1SG.OFOC wrap
‘It was 1the fish that I wrapped.’
(10)
Lan
laa
taxañ?
what 1SG.OFOC wrap
‘What did I wrap?’
On the other hand, benefactive objects contained in a prepositional phrase cannot be
extracted either for questioning (11) or focus (12). Recall that dative objects are not found
in prepositional phrases, only as the applied object in applicatives, which is why there are
no corresponding dative examples.
(11)
(12)
*Ngir kan
laa
taxañ
for
who 1SG.OFOC wrap
‘For whom did I wrap the fish?’
jën
fish
*Ngir Bintou laa
taxañ
jën
for
Bintou 1SG.OFOC wrap
fish
‘It was for Bintou that I wrap the fish.’
wi?
DEF1
wi.
DEF1
Looking now at applicatives, in benefactive and dative applicatives, either object,
theme or applied, can be questioned (13) or focused (14). The benefactive applied object
has been extracted in (13a) and (14a). The theme has been extracted in (13b) and (14b).
Benefactive - questioning applied
(13) a.
Kan laa
taxañ-al
jen
wi?
who 1SG.OFOC wrap-APPL fish DEF1
‘For whom did I wrapped the fish?’
Benefactive - questioning theme
b.
Lan laa
taxañ-al
Bintou?
what 1SG.OFOC wrap-APPL Bintou
‘What did I wrap for Bintou?’
!77
Benefactive - focus applied
(14) a.
Bintou laa
taxañ-al
jen
wi.
Bintou 1SG.OFOC wrap-APPL fish
DEF1
‘It was Bintou for whom I wrapped the fish.’
Benefactive - theme applied
b.
Jën wi
laa
taxañ-al
Bintou.
fish DEF1 1SG.OFOC wrap-APPL Bintou
It was the fish that I wrapped for Bintou.
Extraction in dative applicatives follows the same pattern. Either object can be
questioned (15) or focused (16).
Dative - questioning applied
(15) a.
Kan
laa
jox
neexal?
who
1SG.OFOC give gift
‘To whom did I give a gift.’
Dative - questioning theme
b.
Lan
laa
jox
what
1SG.OFOC
give
‘What did I give to the child?’
xale
child
bi.
DEF1
Dative - focus applied
(16) a.
Xale bi
laa
jox
neexal.
child DEF1 1SG.OFOC give gift
‘It was the child to whom I gave a gift.’
Dative - focus theme
b.
Neexal
laa
jox
xale
gift
1SG.OFOC
give child
‘It was a gift that I gave to the child.’
bi.
DEF1
The data from applicatives indicate that applied objects and direct objects have
the same object status in Wolof. Both objects can be placed as the first object, both can be
!78
extracted for focus or questioning. Since both objects have direct object status,
benefactive and datives applicatives are symmetrical in Wolof.
3.1.4. Object omission
Up to this point, benefactive and dative applicatives have patterned together. The purpose
of this section is to outline a syntactic difference that does exist between the two. In
benefactive applicatives, themes can be omitted without altered verbal morphology but in
dative applicatives, the theme cannot be. It can only be omitted via deletion using a
detransitivizing suffix, -e.2 For the purposes of this discussion, I assume that in
benefactives, object omission is an example of unspecified object deletion.
Unspecified object deletion refers to a process by which (one of) a predicate’s
object(s) is not realized overtly.
(17)
a.
Bill ate ∅ while driving to work.
b.
Bill ate breakfast while driving to work.
In (17a), no direct object is mentioned but it is understood that there was some food item
or meal that Bill ate. I assume null objets are represented in the syntax, following
Roberge and Cummins (2005). This means that I do not take unspecified object deletion
to be a valency-decreasing operation.
In Wolof, many verbs share the property of allowing object omission, which I take
to be unspecified object deletion as in English.
Unspecified object deletion
(18) a.
Dama-y
jàng.
1SG.PFOC-IMP read/study
‘I am reading/studying.’
2
This suffix is not to be confused with the homophonous instrumental/locative applicative morpheme, -e,
which will be discussed in chapter 4.
!79
Theme object realized
b.
Dama-y
jàng
téere
1SG.PFOC-IMP read/study book
‘I am reading/studying the book.’
bi.
DEF1
In benefactive applicatives, theme objects can be omitted without rendering the sentence
ungrammatical. Again, what makes this unspecified object deletion is that there is no
valency decreasing morphology present.
(19)
a.
Daaj-al
naa Tapha ∅.
construct-APPL 1SG Tapha
‘I constructed (something) for Tapha.’
Dative applicatives, on the other hand, are judged as incomplete to the point of being
unacceptable when either the dative or theme object is left unspecified.
(20)
a.
Jox na
ma
xaalis.
give 3SG 1SG.OBJ
money
He/She gave me money.
b.
*Jox
na
xaalis.
give
3SG money
He/She gave money.
c.
*Jox na
ma.
give 3SG 1SG.OBJ
He/She gave me (something).
In order to omit one of the objects of such a verb, one must add a detransitivizing suffix, e, to the verb.
Theme object suppressed
(21) a.
Jox-e
na
ma.
give-DETRAN 3SG 1SG.OBJ
He/She gave me (something).
!80
Dative object supressed
b
Jox-e
na
give-DETRAN 3SG
He/She gave money.
xaalis.
money
Some other verbs that behave the same way are wan ‘to show’, may ‘to offer’, and
yooñee ‘to send’.
As an interesting side note, there is at least one verb that is optionally dative or
transitive, the verb bind ‘to write’. It allows a dative applied object but does not require it
like the dative verbs just mentioned, as shown by (22a) and (b). I note that bind is similar
to other dative verbs because the additional object, Omar in (22a), is assigned a dative
theta role (either source or goal) and there is no overt applicative morphology on the
verb. Interestingly, when bind is used transitively as in (22b), unspecified object deletion
is allowed, as seen in (22c).
(22)
a.
Dama-y
bind Omar bataaxal.
1SG.AFF-IMP write Omar letter
“I am writing Omar a letter. / I wrote a letter to Omar.”
b.
Dama-y
bind
bataaxal.
1SG.AFF-IMP
write letter
“I am writing a letter.”
c.
Damay
bind.
1SG.AFF write
‘I am writing.
Even though unspecified object deletion of the theme is allowed, it is ungrammatical
when there is a dative object present, as illustrated in (23a). In these cases, the theme
object can only be omitted if the detransitivizing morpheme -e is present on the verb like
!81
the other dative verbs. The presence of the dative object turns the verb into a dative and
unspecified object deletion is not longer possible, just as with other dative applicatives.
(23)
a.
*Dama-y
bind
Omar.
1SG.AFF-IMP write-APASS Omar
“I am writing to Omar.”
b.
Dama-y
bind-e
Omar.
1SG.AFF-IMP write-APASS Omar
“I am writing to Omar.”
Optionally ditransitive verbs are not very productive in Wolof. Only the verb bind ‘to
write’ has been identified, to the best of my knowledge, as allowing an optional dative,
although other similar verbs may exist.
In summary, benefactive applicatives allow deletion of the theme object. In dative
applicatives, the theme cannot be deleted via unspecified object deletion. The objects
must be suppressed by a detransitivizing suffix -e. The reason for the difference in
acceptability of unspecified object deletion between the two types of applicatives will be
discussed in section 3.3.2.
3.1.5. Summary of object properties
In benefactive and dative applicatives, both direct and applied objects show similar object
properties. Either can be placed directly after the verb. Both objects are pronominalized
using clitic pronouns and can both be pronominalized simultaneously, a fact that will be
relevant in chapter 4. Finally, either object can be extracted for focus or questioning.
Based on these tests, I conclude benefactive and dative applicatives both involve two true
“direct” objects with equal status. They are therefore both instances of the symmetrical
applicative type.
One difference does exist between the two types of applicatives. Unspecified
object deletion in benefactives is possible with the theme object while dative applicatives
!82
require the suffix -e to suppress either the theme or dative object. This difference is
important because it shows that benefactive and dative applicatives must be different in
some way syntactically. More evidence for separating the analyses of benefactive and
dative applicative will be outlined in section 3.3.2.1.
3.2. C-command
We have just seen that Wolof dative and benefactive applicatives are symmetrical
applicatives differing in object behaviour only for unspecified object deletion.
Nevertheless, we still don’t know the precise nature of the structural relationship between
the two objects (e.g. theme and dative or theme and benefactive). In the literature on
applicatives, the structural relationship between the two objects is a topic of much
interest. There are many tests available to determine the c-command relationship between
objects. The most common ones used in the literature are quantifier binding,
reflexivization, weak cross-over, negative polarity items, superiority effects, and
reciprocal phrases like each...the other. (See Barss and Lasnik 1986 and Larson 1988 for
further discussion of these tests.) Some of these tests are amenable to Wolof but not all.
Quantifier binding, reflexivization and weak cross-over are amenable to Wolof and are
discussed below.
3.2.1. Tests
Before applying quantifier binding, reflexivization and weak cross-over to Wolof
applicatives, I evaluate the validity of these tests in Wolof. In the following subsections, I
first show that these tests do target c-command relationships using transitive sentences
and then I apply the tests to benefactive and dative applicatives.
3.2.1.1. Quantifier Binding
Previous research has established that a quantified applied object can bind an anaphor in
the theme object but not vice versa. Larson (1988) shows this for English (24) and
!83
Marantz (1993) for Swahili (25). In (24a) the applied object is quantified, every worker,
and binds the possessive anaphor in the theme, his. The sentence in (24b) shows that a
quantified theme object, every paycheck, cannot bind the possessive anaphor in the
applied object, its.
(24)
a.
b.
I sent every workeri hisi paycheck.
*I sent itsi owner every paychecki.
The same is true in (25). The quantified applied object in (25a), kila mwandishi ‘each
author’, binds the possessive anaphor chake ‘his’. However, the quantified theme in (25b)
does not bind the anaphor in the applied object, wake ‘its’.
(25)
a.
Ni-li-m-som-e-a
[kila
SP-pst-OP-read-APPL-fv each
“I read each author his book.”
mwandishi]i kitabu
author
book
chakei.
his
b.
*Ni-li-m-som-e-a
[kila kitabu]i mwandishi wakei.
SP-pst-OP-read-APPL-fv each book
author
its
‘*I read its author each book.’
(Marantz 1993:117)
In Swahili, and other Bantu languages, the applied object can scramble to the
right and follow the theme, giving the opposite linear object order: Theme - Applied (see
chapter 2, section 2.1.3). Even though the surface word order has changed, the original
binding relationship is maintained. The quantified applied object still binds the possessive
anaphor in the theme object. The following example shows that the quantified applied
object kila mwandishi ‘each author’ has been placed after the theme, kitabu chake ‘his
book’ yet still binds it.
(26)
Ni-li-m-som-e-a
kitabu
SP-pst-OP-read-APPL-fv book
‘I read each author his book.’
chakei
his
[kila
each
mwandishi]i.
author
(Marantz 1993:117)
!84
Marantz (1993) assumes that it has been right-adjoined to the VP via Heavy NP Shift,
although the specific analysis of the rightward movement is not the focus here. What is
important is that modifying the word order does not allow the theme to c-command the
applied object. Sentences in Swahili with a quantified theme object and a possessed
applied object are not grammatical regardless of the order of the objects. This shows that
binding and by extension, c-command relationships are not determined by linear order of
the objects in Swahili (compare 27 and 25b to 25a).
(27)
*Ni-li-m-som-e-a
mwandishi
SP-pst-OP-read-APPL-fv author
“*I read its author each book.”
wakei
its
[kila
each
kitabu]i.
book
Based on quantifier binding on applicatives, theorists have concluded that the
applied object is generated higher in the VP than the theme object cross-linguistically.
Now let’s see if this generalization holds in Wolof.
The Wolof quantifier used here has the form X bu nekk where X represents the
quantified noun. This quantifier is a distributive universal quantifier with similar
characteristics to every in English as opposed to a collective universal quantifier like all.
For example, subjects with every require third person singular agreement on the verb,
while subjects with all require plural agreement.
(28)
Every boy eats cake.
*Every boy eat cake.
(29)
All boys eat cake.
*All boys eats cake.
In Wolof, when the subject contains X bu nekk, the verb is marked with third
person singular morphology (30), not plural agreement. Additionally, NPs within the bu
nekk construction can occur as possessors (31), indicating they represent individuals and
not a collective entity. The quantifier cannot be the subject of a collective predicate like
!85
daje ‘meet’ (32), again indicating that it does not refer to a collective entity or group but
rather induces a distributive reading. It also cannot be the subject of a verb containing the
suffix -andoo which means roughly ‘together’ (33) (Tamba et al. 2012:920-923).
3SG verbal agreement
(30) Nit
ku
nekk
lekk
person CREL
exist
eat
“Each/every person ate rice.”
Possessor
(31) xaj-u
xale
dog-POSS
child
“every child’s dog”
bu
CREL
na
3SG
ceeb.
rice
nekk
exist
Collective predicate
(32) *Xale bu
nekk
daje
nañu.
child CREL exist
gather
3PL
intended: “*Every child gathered.”
“together”
(33) *Xale bu
nekk
lekk-andoo
CREL
child
exist
eat-together
intended: “*Every child ate together.”
na
3SG
ceeb
rice
bi.
DEF1
Now that the Wolof quantifier used here has been described, I turn back to
quantifier binding in Wolof. Quantifier binding relies on the assumption that quantifiers
which are coreferential with a possessive anaphor must bind said anaphor. Binding
requires the quantifier c-command the anaphor, assuming binding requires c-command
(Reinhart 1976). The c-command requirement is confirmed in Wolof for the universal
quantifier X bu nekk ‘every/each (lit. X that exists)’ by looking at the binding of subjects
and objects in transitive sentences. Since we know independently that subjects ccommand objects in Wolof (Torrence 2012, Dunigan 1994, Russell 2007), if a quantifier
in object position is able to bind an anaphor in subject position, then we can rule out that
!86
binding requires c-command. In Wolof, this result does not arise. As shown in example
(34b), a quantified object cannot bind the anaphor in subject position.
(34)
*Jëkër-ami
nob na
jabar
bu
husband-3SG.POSS love 3SG wife
CREL
intended: ‘Heri husband loves every wifei.’
nekki.
exist
On the other hand, when the quantified expression góor gu nekk ‘every man’ is in
subject position and the possessive anaphor jabar-am ‘his wife’ is in object position, the
sentence is grammatical with a bound interpretation, as seen in example (35a). When the
object jabaram ‘his wife’ appears in a focus position which is higher than a subject as in
(35b), the quantifier no longer binds the object anaphor. The sentence in (35b) means
every man loves some man’s wife. There is only one wife that is loved by all instead of
the bound interpretation where there are multiple wives, each loved by her own husband.
(35)
a.
Góor gu
nekki nob na
man CREL exist eat 3SG.PRF
“Every mani loves hisi wife.”
jabar-ami.
wife-3SG.POSS
b.
Jabar-amj/*i
la
góor [gu] nekki nob.
wife-3SG.POSS 3SG.OFOC man CREL exist love
“It’s hisj/*i wife that every mani loves.”
To ensure that c-command is indeed responsible for the absence of a bound
reading of (34) and (35b), and not due to linear precedence in the sense of Grodzinski and
Reinhart (1993), the quantifier is placed in a position which precedes the anaphor but
does not c-command it in (36). The quantified expression is embedded in a relative
clause, so although the quantifier, jigeen ju nekk ‘every woman’, precedes the anaphor am, it does not c-command it. The binding possibilities in this sentence for the possessive
anaphor -am ‘his/her’ are the subject, góor ‘man’, or a third party not directly referenced
in the sentence. The anaphor këram ‘his/her house’ cannot be bound by the quantified
!87
expression jigeen ju nekk ‘every woman’ even though the quantified expression precedes
the anaphor.
(36)
Góori [gi
jigeen
ju
nekkj nob] dem na
kër-ami/*j/k.
CREL woman CREL
man
exist love go
3SG house-3SG.POSS
“The mani that every womanj loves went to his/her*j/k/i house.”
Based on this evidence, I conclude the lack of a bound interpretation results from the lack
of c-command rather than the lack linear precedence.
The semantic interpretation of X bu nekk reinforces the conclusion that the bound
interpretation of quantified expressions requires c-command. Evidence from semantic
judgements allows us to confirm speakers are not accepting or rejecting sentences based
on other referential possibilities such as a discourse salient third party or another NP in
the sentence. In the following example, the quantifier is contained within an embedded
clause but the sentence has been paired with a context that pushes for a bound
interpretation. The sentence is unacceptable in this context because the quantified object,
ku nekk naw ‘everyone’ does not c-command the anaphor so a bound interpretation is not
possible.
(37)
You passed the day in the village and you are recounting what happened. You
talked to an old woman, a woman selling peanuts, and a woman who every
man admires. The admired woman visited each man that admired her. She
went to each man’s house to say hello and ask about his news.
a.
#Jigéen jii
[k-uj
nekk naw] dem na kër-ami.
woman DEF1 H-COMP exist admire go 3SG house-3SG.POSS
The womani who everyonej admires went to heri /*hisj house.
The interaction between the context and the sentence in (37a) show an embedded
quantifier cannot be interpreted as binding an anaphor in the matrix clause even though
the quantifier precedes it.
!88
Having now presented several lines of evidence that binding involving the
quantifier X bu nekk indeed relies on c-command in Wolof, we know that quantifier
binding can be used to test the structural relationships between objects in applicatives. I
now turn to quantifier binding in benefactive and dative applicatives.
In Wolof applicatives, quantifier binding gives unexpected results. Binding
depends on the linear order of the objects in the sentence. It is not correlated with the
theta role of the object, like in Swahili. Quantified applied objects can bind theme objects
and quantified theme objects can bind applied objects as long as the quantified object is
first. This is surprising since it was just established that quantifier binding in Wolof
doesn’t rely on linear precedence but on c-command.
In (38a) the applied object góor gu nekk ‘every man’ binds the anaphor -am ‘his/
her’. Placing the applied object to the right leads to unavailability of the bound reading as
seen in (38b) for benefactives and (39b) for datives. The (b) sentences are grammatical,
but not on the bound interpretation. The possessive anaphor can only refer to a discourse
referent not mentioned in the sentence.
(38)
(39)
a.
Bind-al
naa
góor gu
nekki bataaxal-ami.
write-APPL 1SG.PRF man CREL exist letter-3SG.POSS
“I wrote hisi letter on behalf of every authori.”
b.
Bind-al
naa
bataaxal-am*i/j góor gu
nekki.
write-APPL 1SG.PRF letter-3SG.POSS man CREL exist
“I wrote hisj/*i letter on behalf of every authori.”
a.
Yóonee naa
góor gu
nekki xaalis-ami.
send
1SG.PRF man CREL exist money-3SG.POSS
“I sent every mani hisi money.”
!89
b.
Yóonee
naa
xaalis-am*i/j
góor gu
send
1SG.PRF money-3SG.POSS man CREL
“I sent hisj/*i money to every mani.”
nekki.
exist
The linear position of the object thus affects the c-command relationship it holds with the
other object.
Looking now at a quantified theme object (underlined in 40 and 41), it can bind
the applied object when the applied object follows the theme. In the following examples,
the quantified theme téere bu nekk ‘every book’ binds the possessive anaphor -am ‘his/
her’ in the benefactive object (40a) and the dative object (41a).
(40)
(41)
a.
Bind-al
naa
téere bu
nekki bindekat-ami.
write-APPL 1SG.PRF book CREL exist author-3SG.POSS
“I wrote every booki for itsi author.”
b.
Bind-al
naa
bindekat-am*i/j
téere bu
nekki.
write-APPL 1SG.PRF author-3SG.POSS book CREL exist
‘I wrote every booki for its*i/j author.’
a.
Yoonee naa
téere bu
nekki bindekat-ami.
send
1SG.PRF book CREL exist writer-3SG.POSS
“I sent every booki to itsi author.”
b.
Yoonee naa
bindekat-am*i/j
send
1SG.PRF writer-3SG.POSS
‘I sent every booki to its*i/j author.’
téere bu
book CREL
nekki.
exist
Thus far, I conclude from quantifier binding data that, in Wolof, the first object (in
a linear sense) c-commands the second object regardless of thematic role. Varying the
word order leads to different c-command relationships and different binding patterns.
Note that this is unlike what Marantz (1993) reports for Swahili, where modifying the
word order does not change the binding relationships.
!90
3.2.1.2. Reflexivization
I now turn to data from reflexivization. Reflexivization is a standard test for c-command
(Larson 1988) because in many languages a reflexive anaphor must be bound by the noun
phrase with which it is coreferential. Wolof has a reflexive pronoun which agrees with the
number and gender of its antecedent. It is a compound anaphor made up of the noun bopp
‘head’ and the appropriate possessive anaphor. The forms of the reflexive pronoun are
given in Table 2.
Table 2: Wolof reflexive pronouns
1SG
sama bopp
2SG
sa bopp
3SG
bopp-am
1PL
2PL
3PL
sunu bopp
seen bopp
seen bopp
Since there is no morphologically “most simple” form from which the others are derived,
I will use the third person singular form boppam as the default form when referring to the
reflexive anaphor. In (42), the reflexive pronoun boppam ‘himself/herself’ is in object
position and is coreferential with the subject pronoun dafay ‘he/she’. Since the bound
reading is available, we can deduce that the subject c-commands the object, as expected.
(42)
Dafa-yi
sang
3SG.PFOC-IMP wash
‘She washes herself.’
boppami.
REFL
If the reflexive anaphor precedes the subject, as in (43a), the sentence is
ungrammatical. The anaphor, sama bopp ‘myself’, is in a topic position and the bound
interpretation of sama bopp is unavailable. On the other hand, in (b), the reflexive
anaphor is in a topic position but this time, a topicalized subject pronoun, man ‘I/me’
precedes it. The reflexive anaphor then can be bound by the higher subject pronoun and
as a result, the sentence is grammatical.
!91
(43)
a.
b.
*Sama
bopp, dem
1SG.POSS
head
go
Myself, I go there...’
naa
1SG
fa...
LOC
Man, sama
bopp, dem naa fa...
1SG
1SG.POSS head go
1SG LOC
Me, myself, I go there...’
The examples in (43) show that the phrase boppam must be bound, which is one criteria
for anaphora, but it does not show which binding domains are possible. Based on the
interpretations available in embedded clauses, (44), the anaphor boppam can only be
bound by an antecedent in the same clause (44a). Binding from the matrix subject is
ungrammatical (44b).
(44)
a.
Wax naa [ni
wat
ngeeni seeni
bopp.]
speak 1SG COMP shave 2PL
2PL.POSS head
✓‘I said that you shaved yourself.’ (reflexive)
✓‘I said that you shaved my head.’ (non-reflexive)
b.
Wax naai [ni
wat
ngeen sama
bopp*i.]
speak 1SG COMP shave 2PL
1SG.POSS head
✗‘I said that you shaved myself.’ (reflexive)
✓‘I said that you shaved my head.’ (non-reflexive)
The sentence in (b) is only grammatical under the non-reflexive reading of boppam, ‘I
said that you shaved my head’. As a reflexive anaphor, boppam must be bound locally.
When there is no appropriate local antecedent, boppam is interpreted as a simple
possessed noun. These data show the reflexive pronoun boppam is an anaphor that is
sensitive to c-command and clausal boundaries. This sensitivity allows us to apply it to
applicative objects to explore the c-command relationship between objects.
In Wolof applicatives, the first object can be the antecedent for a reflexive
anaphor in the second object. This confirms what was seen earlier with quantifier
!92
binding: different word order implies different structural relationships. Since the object
order in benefactive and dative applicatives is variable, either the applied or theme object
can be the antecedent for a reflexive anaphor.
In benefactive applicatives, the reflexive anaphor, boppam, in second position can
be bound by the preceding applied object, Boris as in (45a). When the anaphor, boppam,
appears before the applied object, the sentence is ungrammatical on the bound
interpretation (45b).
(45)
a.
Sang-al
nga
Borisi boppami.
wash-APPL 2SG
Boris REFL
‘You washed himselfi for Borisi.’
‘You washed his*i/j head for Borisi.’
b.
Sang-al
nga
boppam*i/j
Borisi.
REFL
wash-APPL 2SG
Boris
‘*You washed himself*i/j for Borisi.’
‘You washed his*i/j head for Borisi.’
This is true for dative applicatives (46) as well.
(46)
a.
Wan naa Borisi bopp-ami.
show 1SG Boris REFL
‘I showed Borisi himselfi.’
‘I showed his*i/j head to Borisi.’
b.
Wan
naa
bopp-am*i/j Borisi.
show
1SG REFL
Boris
‘*I showed himselfi to Borisi.’
‘I showed his*i/j head to Borisi.’
The binding relationship is lost when the antecedent Boris, be it dative or
benefactive or theme, appears to the right. The bound interpretation of the (b) examples
!93
remains unacceptable regardless of whether Boris is interpreted as the applied object or
the theme. It is unclear what the difference between the two readings would be. Whether
you wash Boris for himself or wash himself for Boris (or show Boris to himself in the
dative example), in both readings, the effect is the same, Boris is washed for Boris. Both
readings are subject to the same linearity effect in which the first object c-commands the
second. Whether the reflexive anaphor in the (b) examples is replacing the theme or the
applied object, the result is always an ungrammatical sentence on the bound
interpretation.
Reflexivization data confirm the quantifier binding data and the conclusion that
the first object c-commands the second regardless of thematic role. The conclusion drawn
from these data is not expected considering previously studied languages with
applicatives like Swahili, discussed in 3.2.1.1. Data from weak cross-over tests show that
in Wolof applicatives, the theme is in fact generated below and within the c-command
domain of the applied object, as in Swahili. The variable word order and c-command
facts will be explained in section 3.3.3.
3.2.1.3. Weak Cross-Over
Another test used to identify c-command relationships between NPs is Weak Cross-Over
(Larson 1988). When a sentence contains a wh-phrase and an anaphor, there are two
possible outcomes for the interpretation. Either a coreferential reading between the whphrase and the anaphor is possible or it is not. The unavailability of a coreferential
reading is said to result from a weak cross-over violation.
A weak cross-over violation arises when the trace of a wh-phrase is c-commanded
by an NP containing a coreferential anaphor. Using this, if a coreferential interpretation
between the wh-phrase and the anaphor is not available, then one knows a weak crossover violation has occurred. The violation indicates that the trace, and by extension the
original position of the wh-phrase, is c-commanded by the anaphor. If a coreferential
!94
reading is available, then we can conclude the anaphor does not c-command the trace of
the wh-phrase.
To see how this works, let’s look at an example from English. The wh-expression
in (20a) is coreferential with the anaphor in the direct object. The wh-expression, which
mother, raises from subject position to CP without ever crossing the anaphor, her, in
object position, as we see in the structure in (47b). The trace of the wh-phrase is not ccommand by the anaphor, her.
(47)
a.
Which mother1 loves her1 child?
b.
On the other hand, the sentence in (48a) is the only acceptable if his refers to a
male not mentioned in the sentence. The coreferential interpretation in (48a) is
unavailable because the wh-expression, which child, raises across the anaphor, his, in
subject position on its way to CP. In the illustrations, movement is tracked with
alphabetic indices and coreference with numerical indices to avoid confusion. In this
case, the trace of the wh-phrase is c-commanded by the anaphor his.
(48)
a.
Which child1 does his*1/2 mother love t ?
!95
b.
Applying this test to applicatives in English, the applied object can be questioned
and fronted when the theme is co-referential (49a). A weak cross-over effect arises when
a theme wh-expression moves over a pronoun in the applied NP (49b) (taken from Larson
1988). This contrast is taken to mean the applied object asymmetrically c-commands the
theme object.
(49) a.
b.
Which mani did you send t hisi paycheck?
Whosei pay did you send his*i/j mother t ?
If Wolof grammar is sensitive to weak cross-over, then we expect to see the same
contrasts observed for English in (47) and (48). This is in fact what occurs in Wolof.
To start, the Wolof sentences in (50) are analogous to the English examples in
(47) and (48).
(50)
a.
Ban yaayi ti mo
nob
which mother
3SG.SFOC love
“Which motheri loves heri child?”
b.
Ban
doomi yaay-am*i/j
mo
nob
which
child
3SG.SFOC
love
mother-3SG.POSS
doom-ami?
child-3SG.POSS
ti ?
!96
Which childi does his*i/j mother love?’
c.
Yaay-am*i/j mo
nob ban
doomi?
mother
3SG.SFOC love which child
“His/her*i/j mother loves which son/daughteri.”
In (50a) the wh-expression, ban yaay, is in subject position and is coreferential with the
anaphor doom-am ‘her child’ in object position. The sentence is acceptable with the
coreferential reading, meaning no weak cross-over violation has occurred. In (50b) the
object wh-expression is questioned and thus appears in a focus position, c-commanding
the anaphor in subject position and the coreferential reading is blocked. This confirms
that the trace of the wh-expression is c-commanded by the the NP containing the anaphor.
In (28c), the anaphor, yaay-am, is in subject position and the in situ wh-phrase, ban
doom, is in object position. In this example, like (50b), the coreferential reading is not
available. Even though the wh-phrase remains in situ overtly, I assume it moves over the
subject to the left periphery at LF to receive its interpretation, leaving a trace behind in
object position following Huang (1982). It is this trace that is c-commanded by the
anaphor, thereby causing the weak cross-over violation and ultimate ungrammaticality of
the coreferential reading.
As expected, a wh-expression that is coreferential with an anaphor cannot be
generated in a position c-commanded by the anaphor in Wolof. These sentences show
weak cross-over does target c-command in Wolof and can be applied to other structures
like applicatives. By placing an anaphor and a wh-expression in each object position,
weak cross-over identifies which object is generated in the c-commanding position and
which in the lower, c-commanded position.
As I show below, weak cross-over data from Wolof applicatives show the applied
object in benefactive and dative applicatives c-commands the theme and it is not possible
for the theme to c-command the applied object. Only the applied object can be questioned
while still being coreferential with the anaphor in the theme object position. Unlike
!97
quantifier binding and reflexivization, the linear order of objects is irrelevant since only
one object appears post-verbally, and one in the left periphery.
Coreference possibilities and weak cross-over effects are shown for a dative
applicative in (51). When the wh-phrase is in the applied object, gan góor ‘which man’ as
in (51a), no weak cross-over violation occurs. Coreference between the two objects is
available. The (b) sentence, however, shows a weak cross-over violation that blocks
coreference when the wh-phrase is the theme object, bataaxalu kan ‘whose letter’.
Applied object questioned
(51) a.
Gan
góori nga
yónnee ti bataaxal-ami?
which man
2SG send
letter-3SG.POSS
“Which mani did you send hisi letter?”
Theme object questioned
b.
Bataaxal-u kani nga yónnee bindekat-am*i/j ti ?
letter-CS
who 2SG send
author-3SG.POSS
Whosei letter did you send to its*i/j author?’
I conclude that the applied object, gan góor, does not cross over the theme, bataaxalam as
it raises to the left in (51a). In (51b) however, we know the unacceptability is due to weak
cross-over because the sentence is acceptable with a non-coreferential reading. This
means the theme, bataaxalu kan ‘whose letter’, must cross over the applied dative object
when it raises to the left, leaving its trace c-commanded by the anaphor in applied object,
bindekatam ‘its author’. Unlike quantifier binding and reflexivization, these weak crossover data show the theme object is always generated below the applied object, as argued
by previous analyses (Marantz 1993; Ngonyani 1996, 1998; Pylkkänen 2008).
Example (52) shows similar facts for benefactive applicatives. First, let’s look at
example (a). In (52a) the benefactive wh-object (underlined) has raised to the left
periphery and the sentence is acceptable with the coreferential interpretation between the
!98
wh-phrase, ban jigéen ‘which woman’, and the possessive anaphor in the theme object,
xaram ‘her sheep’. This means, as with the dative example, the anaphor xar-am doesn’t
c-command the trace of the benefactive applied wh-phrase. Looking at the sentence in
(b), where the wh-phrase appears as the theme object, coreference is disallowed due to a
weak cross-over violation. After the theme object, bataaxalu kan ‘whose letter,’ raises to
the left, its trace is c-commanded by the anaphor in the applied object, and the
coreference is blocked. From this we can deduce that the wh-phrase was generated within
the c-command domain of the applied object.
(52)
a.
Ban
jigéeni nga
rey-al
ti xar-ami?
which woman 2SG.OFOC kill-APPL
sheep-3SG.POSS
“For which womani did you kill heri sheep?
b.
Xar-u
kani nga
rey-al
borom-ami/*i
t?
sheep-CS who 2SG kill-APPL owner-3SG.POSS
“Whose sheepi did you kill for his/her*i/j owner?
In contrast, when a benefactive object is contained in a prepositional phrase, the
wh-phrase theme can be coreferential with an anaphor in the oblique benefactive object.
In this construction, the theme is generated outside the c-command domain of the oblique
benefactive object. In (53) the theme wh-phrase xaru kan ‘whose sheep’ binds the
anaphor -am attached to borom, the benefactive object. Since coreference is possible, we
know the anaphor -am doesn’t c-command the trace of the wh-phrase.
(53)
a.
Xar-u
kani nga rey t ngir borom-ami?
sheep-CS who 2SG kill
for
owner-3SG.POSS
“Whose sheepi did you kill for itsi owner?
!99
b.
The weak cross-over facts presented in this section indicate that in applicatives,
the applied object asymmetrically c-commands the theme object and I conclude that it is
consistently generated in the higher position.
3.2.1.4. Summary of c-command tests
The c-command tests used in this section give unexpected results. According to quantifier
binding and reflexivization, the theme can c-command the applied object or vice versa.
However, weak cross-over facts indicate that only the applied object c-commands the
theme object. Why the theme object can c-command the applied object in quantifier
binding and reflexivization tests but not in weak cross-over tests will be addressed in
detail in section 3.3.3. In the next section of this chapter, I argue the applied object is
always generated above the theme object. Instances of the theme c-commanding the
applied object are due to scrambling of the theme object and do not reflect the merge
positions of the objects.
3.3. Analysis
In this section, I show that two types of Appl functional heads account for syntactic
properties of benefactive and dative applicatives. As stated in chapter 2, I do not adopt
High/Low Appl structures of Pylkkänen (2008) for benefactive and dative applicatives.
Instead, I adopt the proposal put forth in Georgala (2012) in which there are two types of
!100
applicatives: thematic and raising.3 The differences in thematic and raising applicative
types arise from selectional differences of the predicates they merge with. I argue
benefactive applicatives in Wolof are of the thematic type. Benefactive applied objects
are selected by ApplT, above the minimal VP. Dative applicatives, on the other hand, are
raising applicatives. Dative objects are selected by the verb and merge within the minimal
VP. The dative object then raises to the ApplE projection for licensing. Once raised, the
dative object is in the same structural position as the benefactive applied object, which
explains the syntactic similarities between the two. I explain the two types below in more
detail.
3.3.1. Benefactives
Recall from chapter 2 that thematic applicatives involve a predicate (VP) and a functional
head called ApplT. Recall also from Chapter 2, that the benefactive object is not a
defining part of the event (Marantz 1993). This translates to it merging outside of the
event syntactically, with the event as the semantic equivalent of the VP. ApplT merges
with VP and establishes a relationship between its object, the benefactive, and the event,
the VP.
(54)
3
The third applicative structure sketched out in chapter 2 will be addressed in detail in chapter 4.
!101
Case assignment is straightforward using the current Minimalist approach to Case
assignment, whereby it is assigned via functional heads only. ApplT assigns Case to the
object in its domain, the theme object, and v assigns case to the object in its domain, the
benefactive object.
The analysis presented here for benefactives does not differ significantly from
previous analyses of benefactive applicatives. My analysis is adopted directly from
Georgala (2012) and the data from benefactive applicatives in Wolof does not necessitate
many changes to the theory.
I now provide a complete derivation of a benefactive applicative using ApplT. The
sentence in (55a) shows a benefactive applicative and its syntactic representation is given
in (55b). The first step in the derivation of such a sentence is the merging of theme object
and the verb which compositionally defines the event (Marantz 1993). ApplT then merges
with the VP via Event Identification (Pylkkänen 2008). The ApplT head takes the VP as
its complement, selects for the benefactive object Omar, and relates the object and the
event to each other. This allows the speaker and listener to identify the particular
occurrence of the event, in this case, writing a letter. In this sentence we can exclude
other occurrences of letter writing that are not for the benefit of Omar. The verb then
raises to Appl and -al attaches to the verb. Once this is complete, the agent is added to the
structure by another syntactic head in the extended verbal projection, v.4 The default
word order, that of Verb - Benefactive - Theme, obtains when the verb raises from ApplT
to v.
(55)
4
a.
Damay bind-al
Omar bataaxal.
1SG.AFF write-APPL Omar letter
‘I wrote a letter on behalf of Omar.’
Kratzer’s Voice (1996) is also a candidate for this job. The choice of Voice or v does not affect the analysis
here and the use of v in this thesis is arbitrary.
!102
b.
This analysis also accounts for the compatibility of benefactive applicatives with
both transitive and intransitive predicates. Since ApplT takes an event as its complement,
it is not sensitive to the transitivity of said predicate. A benefactive applicative used with
an intransitive verb is given in (56a) and its representation in (56b).
(56)
a.
Dafa-y
wax-al
Boris.
3SG.PFOC-IMP speak-APPL Boris
‘He/she is speaking for/on behalf of Boris.’
b.
A thematic applicative approach correctly accounts for the fact that the benefactive object
c-commands and precedes the theme object in ditransitive applicatives (those involving
two objects). It also accounts for the fact that a benefactive object is compatible with an
!103
intransitive predicate. By selecting a VP as its complement, ApplT is insensitive to the
presence of additional objects.
3.3.2. Datives
I now turn to the structure of dative applicatives. Data from word order and binding show
that dative applicatives have a lot in common with benefactive applicatives. Nevertheless,
there is one important difference between the two: unspecified object deletion as
presented in 3.1.4. In this section, I argue that datives are derived differently than
benefactives and are better analyzed as raising applicatives (Georgala 2012). The raising
applicative analysis is advantageous because it accounts for the selectional differences
between the two types - benefactive applied objects are optional while datives are not while still accounting for their structural similarities (e.g. c-command).
Recall from chapter 2 that raising applicatives result from a ditransitive verb that
merges with an expletive Appl (ApplE), allowing the dative object to raise out of the VP
to the specifier of ApplE (Georgala 2012). This Appl is an expletive head because it does
not select an argument and has no semantic content. The relationship the dative object
plays in the event is encoded by the verb.
(57)
!104
The dative object, then, is not really an applied object in the same sense as a benefactive
object because it is selected by the verb and not a functional head. However, it shares
properties with applied objects like benefactives because it occupies the same position,
the specifier of an Appl phrase merged with the VP. The dative object raises to the
specifier of Appl to check the EPP feature on ApplE and receive Case. Assuming Case is
assigned via Agree with functional heads to NPs within their domain, the dative object
receives Case from v once in [Spec,ApplP] and the theme receives Case from ApplE.
Let us see how this analysis works in Wolof dative applicatives using the sentence
in (48). The ditransitive verb wan ‘to show’ selects two objects. First, it merges with the
theme object natal yi ‘the photos’. Then the VP merges with the dative object, Issa. Then
the ApplE head merges. ApplE has an EPP feature that needs to be satisfied either by
Merge or Move (see McGinnis 2001, 2003 for arguments in favor of an EPP feature on
Appl). The dative object moves from the VP to the specifier of the ApplE phrase and
checks the EPP feature of ApplE. The movement of the dative object produces a structure
that parallels the benefactive applicative structure. The structure is illustrated in (58b).5
(58)
a.
Moustapha wan na
Issa natal
Moustapha show 3SG Issa photo
‘Moustapha showed Issa the photos.’
yi.
DEF.PL
The verb movement and subject movement to higher projections are not shown in the tree for simplicity.
The final word order see in (48a) obtains once these movements take place.
5
!105
b.
Although some dative verbs are not obligatorily ditransitive, like bind ‘to write’, I
adopt the same structure for all ditransitive verbs with dative objects. The sentence in
(59a) below is formed via the same process. The verb merges with the theme object,
bataaxal ‘a letter’, then merges with the dative object. Like the wan example in (58), the
verb selects both objects. Then the Appl head merges with an EPP feature and the dative
object moves to its specifier to check the EPP feature and be licensed.
(59)
a.
Damay
bind Omar bataaxal.
1SG.AFF write Omar letter
‘I wrote Omar a letter. / I wrote a letter to Omar.’
The transfer of possession relationship between bataaxal ‘letter’ and Omar is encoded by
the verb bind ‘to write’ since ApplE has no semantic content.
One motivation for the presence of ApplE in dative applicatives is Case
assignment. Without the Appl head, one must assume the verb assigns Case to one or
both of the objects. This contradicts current assumptions about the role of functional
heads in Case assignment in the Minimalist framework, adopted here. Case is assigned
via functional heads to NPs within their domain. In simple transitive sentences, the object
receives case from v. In ditransitive applicative sentences, v assigns case to the dative
!106
object but ApplE is needed to assign case to the theme object. Case assignment is
indicated with arrows.
(60)
A second motivation for ApplE in dative applicative structures is to maintain
continuity with analogous sentences in other languages, like Greek, that also have
movement of the dative object out of the VP to a higher functional projection. Quantifier
float in Greek confirms the movement of the dative object from [Spec,VP] to a higher
position within the extended verbal phrase as shown in (61) (repeated here from example
(84) in Chapter 2).
Floating quantifier - Dative applicative
(61) Servira
tus
pelates
olus
served.1SG the.ACC customers.ACC all.ACC
‘I served all customers breakfast.’
proino.
breakfast.ACC
(Georgala 2012:127)
Notice in this sentence that the quantifier olus ‘all’ appears to the right of the dative
object tus pelates ‘the customers’ even though olus scopes over tus pelates. This shows
the dative object was originally generated to the right of the quantifier but has moved to a
higher position while leaving olus behind. Georgala argues the dative object has moved
!107
to the specifier of the ApplE projection. Analogous examples are not available for Wolof,
but I assume the same structures.
Finally, I assume that datives involve the ApplE head because there is evidence for a
functional projection between the VP and the vP in Wolof. The theme object, bataaxal bi,
can appear before the dative object, Omar, but still after the verb, bind (62).
(62)
Dama-y
bind
bataaxal
1SG-IMP write
letter
‘I wrote the letter to Omar.’
bi
DEF1
Omar.
Omar
There must be some projection above VP but below vP, that hosts this movement. The
ApplE phrase is the natural choice for this functional projection. The combination of
evidence from Case assignment, cross-linguistic comparison, and variable word order
support the presence of ApplE in the structure of Wolof dative applicatives.
Now that I have shown the need for ApplE in dative applicatives, I show that the
dative object must be generated in VP rather than simply merging in ApplE. In other
words, dative applicatives cannot be analyzed successfully as thematic applicatives.
Generating the dative object within the minimal VP in raising applicatives, is supported
by the ungrammaticality of omitted theme objects. I assume the lack of unspecified object
deletion in dative applicatives results from the transfer of possession relationship between
the two objects which is encoded by the verb. The theme object is transferred to the
dative object. Not specifying one of the objects involved in the transfer of possession is
unacceptable because the transfer of possession fails, either there is nothing to be
transferred or no one to receive it. Either way, a semantic condition of the verb is violated
and the sentences are judged to be unacceptable.
Unspecified object deletion of dative object
(63)
a.
*Moustapha wan
na
natal
yi.
Moustapha
show 3SG photo DEF.PL
‘Moustapha showed (someone) the photos.’
!108
Unspecified object deletion of theme object
b.
*Moustapha wan na
Issa.
Moustapha show 3SG Issa.
‘Moustapha showed Issa (something).’
No unspecified object deletion
c.
Moustapha wan na
Issa natal
Moustapha show 3SG Issa photo
‘Moustapha showed Issa the photos.’
yi.
DEF.PL
The need to specify both objects is seen in English as well, where dative applicatives also
resist transitive usage.
(64)
a.
I gave Sally a present.
b.
*I gave Sally ∅.
c.
*I gave ∅ a present.
If the dative object were selected by and related to the event by a functional head,
say ApplT, then the unacceptability of unspecified object deletion is mysterious in (63)
and (64). Remember that ApplT does not create a relationship between the applied and
theme object. It merely associates the applied object with the event. In the case of
benefactives, the theme does not need to be explicit for such an association to occur.
Since benefactives are not related to the theme object, so leaving the theme unspecified is
grammatical.
(65)
a.
Daaj-al
naa Tapha ∅.
construct-APPL 1SG Tapha
I constructed (something) for Tapha.
As a second argument in favour of distinguishing benefactives from datives in
terms of the applicative head, benefactive and dative applicatives have different
morphology. The morphology of benefactive and datives is determined by the Appl head.
!109
ApplT is realized by the morpheme -al which is added to the verb via head movement of
the verb. ApplE has a null morpheme so no additional suffix appears in dative
applicatives.
The quantifier binding, reflexivization, and weak cross-over facts presented in
3.2. fall out naturally from the thematic and raising applicative structures illustrated in
(54) and (57). The applied object, be it benefactive or a raised dative, asymmetrically ccommands the theme, which allows it to bind the theme. A quantified applied object can
therefore bind an anaphor in the theme object. The applied object can also be the
antecedent for a reflexive anaphor in the theme object. Weak cross-over violations are the
result of wh-movement by the theme over the applied object. I now turn to an explanation
of how variable word order in applicatives is achieved within the Thematic/Raising Appl
analysis.
3.3.3. Variable Word Order
In this section, I argue that the variable word order, in which the theme precedes the
benefactive or dative object, is the result of A-scrambling of the theme to a higher
position within the ApplT or ApplE projection. Recall that when the theme precedes the
applied object, the former c-commands the latter, as shown by quantifier binding and
reflexivization facts (3.2.2.1 and 3.2.2.2). Therefore, the theme must have moved via Amovement to a higher projection in order for binding to be affected, as opposed to the
benefactive object moving rightward to a VP adjunct position, as proposed for Swahili by
Marantz (1993).
In order to analyze object scrambling, I draw on work on passivization in
symmetrical and asymmetrical languages. McGinnis (2001, 2003) argues that the analysis
of symmetrical and asymmetrical passives relies on the notions of an Escape Hatch and
Phase Theory. She proposes that High Appl (ApplT in the terminology adopted here) is a
phase head along with v and C. As a phase head, ApplH has an EPP feature. In fact, it can
!110
have multiple EPP features associated with it and this results in object scrambling in
Wolof and the symmetrical/asymmetrical passive behavior seen in Bantu applicatives.
McGinnis claims that asymmetrical passives involve only one EPP feature
associated with ApplH. ApplH with one EPP feature only has one specifier position. This
position is filled and the EPP feature is checked when the benefactive argument merges.
Assuming Merge occurs before Move (Chomsky 1995, 2000), the theme object cannot
move to check the EPP feature because Merge of the applied object will have already
checked the feature.
Since ApplH constitutes a phase head, passive v can only “see” NPs that are at the
edge of ApplH, with the edge being informally defined as the specifier positions. The
applied object is available to the v probe because it is at the edge of ApplH phase. It can
raise to the specifier of vP and from there become the subject of a passive sentence. The
theme object, however, cannot raise to vP since it is contained in the domain of the lower
ApplH phase which is closed to v.
(66)
Symmetrical passives, in which either object can become the subject of a passive
sentence, arise when ApplH has two EPP features. Two EPP features allow both objects
to occupy specifier positions in the ApplH projection, placing both objects at the edge of
the phase. First, the benefactive merges and checks off the first EPP feature. Then the
theme moves to ApplH and checks the second EPP feature.
!111
Since both objects are in the specifier, they are both available goals for the v
probe. Since the theme is above the benefactive, it represents the closest target for the
passive v probe. The theme thus moves to [Spec,vP] and from there on to the subject
position. The benefactive remains in situ in this case because the EPP feature on v is
checked by the theme. The structure up to vP is shown in (67).
(67)
In languages with symmetrical passives, one or two EPP features on Appl are equally
possible. When one EPP feature is present, the applied object becomes the subject; when
two EPP features are present, the theme becomes the subject.
Now let us consider how McGinnis’ analysis of symmetrical passives can help
explain object scrambling in Wolof. The benefactive - theme word order results when
ApplT has one EPP feature, which is satisfied when the benefactive object merges with it.
(68)
!112
At this point, ApplT has no more EPP features to be checked so the derivation continues
with the merge of v. This means that the benefactive precedes and c-commands the theme
object.
The theme - benefactive word order results when ApplT has two EPP features.
The first feature is checked when the benefactive argument merges but there is still an
EPP feature that needs to be checked.
(69)
The only available goal that can check the second EPP feature is the theme object. It
moves into a second specifier position of the ApplT projection and checks off the
remaining EPP feature.6
(70)
6
No ‘Tucking-in’ (Richards 1999) assumed here.
!113
This places the theme above the benefactive object. In this configuration, the theme now
surfaces as the first object and asymmetrically c-commands the applied object. Binding
from the theme to the applied object is now possible, as we saw in sections 3.2.2.1. and
3.2.2.2.
The movement of the theme object also provides an explanation for the weak
cross-over violations seen when a theme is questioned (section 3.2.2.3). Recall that a
weak cross-over violation occurs when the trace of a wh-expression is c-command by a
coreferential anaphor. Even if the wh-theme A-scrambles to the ApplT or ApplE phrase
before raising to CP, it still leaves a trace in its merge position as the complement of the
verb. This trace is c-commanded by the anaphor in the applied object, as shown in (71),
leading to a weak cross-over violation.
When the applied object contains the wh-phrase and the anaphor is in the theme
object, scrambling of the theme still incurs a weak cross-over violation. The wh-phrase in
the applied object raises at LF for interpretation, which means its trace is in the ccommand domain of the scrambled theme object. The LF positions of the objects are
given in the simplified tree in (71).
(71)
!114
I argue that when the applied object is questioned and the theme object has the
anaphor, the only grammatical derivation is the one without object scrambling. The theme
never leaves the complement of V so it never c-commands the trace of the wh-applied
object. The illustration in (72a) shows this derivation. To contrast, (72b) shows a
derivation involving scrambling of the theme to ApplTP before the wh-benefactive moves
to CP. Since there is a grammatical form, we never see the weak cross-over effects with a
questioned benefactive that arise from a structure like (b).
(72)
a.
!115
b.
The structure for dative applicatives is more or less the same except using ApplE
instead of ApplT and the extra step of the dative object raising to ApplEP, and so datives
show the same behaviour vis-à-vis weak cross-over.
As for quantifier binding and reflexivization, even if scrambling has occurred, the
theme’s trace does not pose a problem because traces do not affect binding relationships.
Binding is determined by the final c-command relation between the theme object and the
anaphor. So in the quantifier binding and reflexivization examples, the scrambled theme
c-commands the applied object and thus can bind it, regardless of the fact that its trace is
in the c-command domain of the applied object.
3.3.3.1. Object scrambling
A final note on object scrambling in applicatives is in order here. The underlying reason
for the scrambling in Wolof is unclear at this point. Object scrambling in other languages
has been correlated with specificity (Diesing 1992, Diesing and Jelinek 1993, Thráinsson
2007). Preliminary data on definiteness and specificity from my semantic and pragmatic
field research on Wolof applicatives are inconclusive. Looking at the literature, I am
!116
unaware of any articles that treat object scrambling in Wolof, other than Schwartz (1975),
who argues that definiteness determines word order.
Schwartz claims that definite objects must be adjacent to the verb and the
definiteness determines which object is the direct object and which is the indirect object.
According to Schwartz, whichever object is definite appears first and is interpreted as the
applied object. In his examples, cited here as (73), the object interpreted as the dative
object is the object that is marked definite, xale bi ‘the child’ in (73a) and muus mi ‘the
cat’ in (73b) (see Schwartz 1975:223 for a more detailed explanation). Note that these
examples do not constitute a case of object scrambling because of the change in meaning
of the two sentences.
(73)
a.
góór gi
jox
në
xale bi
muus.
man DEF1 give 3SG child DEF1 cat
‘the man gave a cat to the child.’
*‘the man gave a child to the cat.’
b.
góór gi
jox në
muus mi
xale.
man DEF1 give 3SG cat
DEF1 child
‘the man gave a child to the cat.’
*‘the man gave the cat to a child.’
The speakers consulted for this thesis, however, do not accept the interpretations given by
Schwartz in example (73). They interpret (a) as given but (b) is interpreted as “The man
gave a child the cat. / The man gave a cat to a child.” The interpretation “The child gave
a child to the cat” is rejected. Even though muus mi ‘the cat’ is in first object position and
definite, Saint Louis speakers still interpret it as the theme.
There are several other problems with Schwartz’ conclusion. The first problem is
that he not only alternated which object was definite in his examples, he alternated the
word order as well. Since there are two variables present in his examples, one cannot be
sure if the theta role assignment is based on definiteness or position. One would need to
!117
know what interpretations are available for (74a) and (74b), which are not included in
Schwartz’ article, in order to determine if the interpretation of being the ‘applied
object’ (the indirect object in Schwartz’ terminology) is truly dependent on definiteness.
(74)
a.
góór gi
jox
në
muus
man DEF1 give 3SG cat
‘The man gave the child a cat.’
xale bi.
child DEF1
b.
góór gi
jox
në
xale muus mi.
man DEF1 give 3SG child cat DEF1
‘The man gave a child the cat.’
My consultants found these sentences to be acceptable. The only interpretation possible
for (74a) is ‘The man gave the child a cat.’ The only interpretation for the available for
(b) is ‘The man gave a child the cat.’ In both examples, xale (bi) ‘child’ is interpreted as
the dative object and muss (mi) ‘cat’ is the theme object. These examples show that the
applied object can be indefinite and it can follow the theme.
According to judgments from the Saint Louisien consultants, the thematic
interpretation seems more sensitive to ‘humanness’ than definiteness. In (73b) the object
muus mi ‘the cat’ is definite and the object xale ‘child’ is not definite. Despite this, xale
‘child’ is interpreted as the recipient and instead of muus mi ‘the cat’. Speakers did not
accept the interpretation in which a child was given to the cat because xale ‘child’ is
human while muus mi ‘the cat’ is an animal and it is pragmatically odd at best to give a
child to a cat.
Evidence from speakers consulted for this dissertation do support that definiteness
does play a role in applicative formation, just not in the exact way Schwartz proposed. It
cannot be the case that both objects are indefinite and unspecified as shown in example
(75).
!118
(75)
a.
*Góór gi
jox
në
muus
man
DEF1
give
3SG
cat
Intended: ‘The man gave a child to a cat.’
xale.
child
b.
*Góór gi
jox
në
xale
man
DEF1
give
3SG
child
Intended: ‘The man gave a cat to a child.’
muus.
cat
From these examples, I conclude that one of the objects must be definite. However, given
the definite object can be located in either first or second object position (see 73 and 74),
it is not possible to maintain the hypothesis that scrambling is directly related to
definiteness. If it were, we would expect the definite object to appear in the same position
in varying sentences but this is not the case, as seen in (76).
(76)
a.
Damay
jox xale
1SG.AFF give child
“I gave the child a gift.”
bi
neexal.
DEF1 gift
b.
Damay
jox
neexal xale bi.
1SG.AFF
give
gift
child DEF.PL
“I gave the child a gift.”
The analysis presented in this chapter correctly captures syntactic processes seen in the
data from Wolof. The role that definiteness and specificity play in object scrambling
requires much additional research before any conclusions can be drawn.
In the rest of this section, I address other approaches to analyzing alternating word
order of objects and show they are not able to correctly account for the Wolof data.
3.3.3.2. Different merges
An alternative approach to variable word order is allowing the objects to merge in
different orders, e.g. the theme could merge with the verb or the applied object could
!119
merge with the verb, rather than the object scrambling. When the theme merges first, it
surfaces in the second object position and is c-commanded by the applied object. If the
applied object merges with the verb first, it surfaces as the second object and is ccommanded by the theme. However, weak cross-over indicates the variable merge
hypothesis is not plausible in Wolof applicatives.
If the arguments could be merged in either order, then we would expect no weak
cross-over effects in applicatives. If the theme merged first, then it would be lower than
the applied object. From such a structure the applied object could be questioned without
incurring a weak cross-over violation (77).
(77)
If the applied object could merge with the verb, the theme object would be in the higher
position. The theme would be allowed to be questioned without incurring a weak crossover violation.
!120
(78)
Such sentences however, are ungrammatical due to weak cross-over as already pointed
out in 3.1.2.3 (examples repeated from 51 above).
Applied object questioned
(80) a.
Gan
góori nga
yónnee ti bataaxal-ami?
which man
2SG send
letter-3SG.POSS
“Which mani did you send hisi letter?”
Theme object questioned
b.
Bataaxal-u kani nga yónnee bindekat-am*i/j ti ?
letter-CS
who 2SG send
author-3SG.POSS
Whosei letter did you send to its*i/j author?’
Variable word order in Wolof therefore can’t arise from variable Merge orders. The
combined data from all three c-command tests show the objects in applicatives have a
fixed merge order: theme before applied. I conclude that the variable binding is better
accounted for via object scrambling.
!121
3.3.3.3. Rightward Adjunction
A second proposal for variable word order in applicatives comes from work on Bantu
languages. Binding evidence in these languages, like Swahili, points to the applied object
moving rightward to an adjunct position which still c-commands the theme object. The
variable word order of benefactive applicatives is argued to be the result of rightward
adjunction.
(81)
In Bantu languages variable c-command does not obtain like it does in Wolof. For
example, in Swahili, when the applied object follows the theme object, it still binds an
anaphor in theme position. The quantified applied object, kila mwandishi ‘each author’
still binds the anaphor in the preceding theme object, kitabu chake ‘his book’.
(82)
Ni-li-m-som-e-a
kitabu chakei [kila
SP-pst-OP-read-APPL-fv
book
his
each
“I read each author his book.” (Marantz 1993:117)
mwandishi]i.
author
In Swahili, the linear position of the object does not affect the binding possibilities of the
objects like it does in Wolof.
!122
Additionally in Swahili, a quantified theme can never bind an anaphor in the
applied object regardless of word order unlike Wolof (example cited from Marantz
1993:117).
(83)
a.
*Ni-li-m-som-e-a
[kila kitabu]i mwandishi wakei.
SP-pst-OP-read-APPL-fv each book
author
its
“*I read its author each book.”
b.
*Ni-li-m-som-e-a
mwandish wakei [kila kitabu]i.
iauthor
SP-pst-OP-readits
each book
APPL
-fv
“*I read its author each book.”
We know from these data that theme objects in Swahili cannot c-command the applied
object. Thematic role of the object seems a more relevant determiner of binding and ccommand than linear position in Swahili. I conclude that variable word order in Swahili
and variable word order in Wolof result from different processes as rightward adjunction
does not account for the c-command facts found in Wolof.
3.3.4. A ‘small clause’ problem
Before ending the discussion on dative and benefactive applicatives, I would like to
briefly look at one other type of analysis of applicatives that has been proposed in the
literature. Several linguists have proposed using a small clause structure embedded within
the minimal VP to account for applicatives (Harley 2002; Beck and Johnson 2004). I also
include Pylkkänen’s Low Appl (2002, 2008) in this section because the structure of Low
Appl is essentially a type of small clause. In this section, I illustrate the problems
associated with adopting such an analysis for Wolof applicatives and show that the
thematic and raising applicative analysis is more fruitful for Wolof.
An applicative formed via a small clause has the following structure. The label of
the X head varies across analyses but the small clause structures proposed are parallel.
!123
For example, Pylkkänen (2008) calls XP Low ApplP, while Harley (2002) calls it PHAVE.
The label is not important here, it’s the structure that is relevant.
(84)
The first problem for a small clause analysis is the compatibility of benefactive
applicatives with intransitive predicates. Whether headed by Low Appl or some type of
null preposition like PHAVE, a small clause relates two entities to each other, either an
object to an object or an object to a property. This means that in order to be licit in
applicatives, there must be two objects present. Benefactive applicatives productively
apply to intransitive predicates, leading to a sentence with only one object. This is one
line of evidence that benefactive applicatives cannot involve a small clause. One could
argue intransitive verbs with a benefactive applicative could be the result of unspecified
object deletion or a null anaphor in theme position. However, this assertion cannot be
maintained in the face of unaccusative and unergative examples like dem ‘to go’ or wax
‘to speak’.
(85)
a.
Wax-al
naa Boris.
speak-APPL 1SG Boris
‘I spoke for/on behalf of Boris.’
b.
Dem-al
naa fa
Boris.
speak-APPL 1SG LOC Boris
‘I went there for/on behalf of Boris.’
!124
One cannot argue that applicatives involving these verbs are the result of unspecified
object deletion or a null anaphor because unaccusative verbs don’t select a theme object7
in any circumstances yet are still compatible with benefactive applicatives.
As for dative applicatives that do require two objects, Larson (2010) highlights a
problem with Pylkkänen’s Low Applicative analysis. According to Pylkkänen, the
recipient (dative object) is not semantically related to the verb. She explains, “Low
applied arguments bear no semantic relation to the verb whatsoever: they bear only a
transfer-of-possession relation to the direct object” (2008:14). Larson points out that the
semantic definition of Low Appl leads to vacuous quantification and false entailments.
The following sentences to show the semantic analysis of Low Appl leads to entailments
which do not hold. The entailment in (86b) is not true of the applicative sentence in (86a).
(86)
a.
John wrote Mary that letter.
b.
John wrote that letter and Bill gave Mary that letter.
Under Pylkkänen, (b) should be an entailment of (a), given that the low applied object
(Mary in this case) bears no relation to the verb. All that matter is that Mary gets the
letter. However, (b) is not an entailment of (a). Larson argues that Pylkkänen's departure
from the Neo-Davidsonian analysis of applicatives results in the entailment problems.
Low Appl is therefore not a tenable approach to dative applicatives.
Aside from the problem with the semantics of Low Appl brought forth by Larson,
the lack of other types of small clauses like resultatives and to-datives in Wolof are
problematic for any small clause analysis. Snyder (2001) and Beck and Johnson (2004)
argue the presence of resultatives clauses and to-datives correlates with the presence of
double object constructions (applicatives) in a given language, since they involve the
7
The term ‘theme object’ used here is not to be confused with ‘theme argument’. Unaccusative verbs do
select a theme argument, which appears in subject position, but they select no additional arguments that
could appear in object position.
!125
same small clause structure. Conversely, if a language does not have resultative or todatives, it is expected not to allow applicatives.
Synder (2001) proposes a principle, which he calls Principle R, which facilitates
the combination of a small clause with an event of a different semantic type. Assuming
resultatives and double object constructions (applicatives) both have the same structure in
the syntax, the small clause, both require the presence of Principle R in the language in
order to be felicitous. So if a language has resultatives then it should also allow double
object constructions (applicatives) and vice-versa. Conversely, if a language does not
have resultatives, or any other type of small clause construction, then it is expected not to
allow double object constructions.
Preliminary data suggest a total absence of resultatives and to-datives in Wolof.
First, resulting states in Wolof cannot be expressed using a small clause, as seen in (87a)
unlike its equivalent in English. The adjective ubbiku ‘open’ must be contained in a
relative clause with an overt complementizer, mu, shown in (87b).
Resultative Small clause
(87) a.
*Bay
naa
bunt
leave
1SG
door
‘I left the door open.’
bi
ubbiku.
DEF1 open
Relative clause
b.
Bay
naa bunt
bi
mu
leave 1SG door DEF1 CREL
‘I left the door open.’
ubbiku.
open
Second, dative objects cannot be expressed in a prepositional phrase. The
preposition ci which roughly translates to to in English is used with locative objects not
recipients like Omar in (88a). Instead, to express dative objects, an applicative structure
is used as shown in (b).
!126
to-Dative
(88) a.
*Yonnee naa
bataaxal
send
1SG letter
‘I sent a letter to Omar.’
ci
to
Omar.
Omar
Dative applicative
b.
Yonnee naa Omar bataaxal.
send
1SG Omar letter
‘I sent a letter to Omar. (I sent Omar a letter.)’
I thus conclude that complex predicates with embedded small clauses are not
possible in Wolof. Resulting states in Wolof are expressed using relative clauses as in
(87b). Prepositional datives do not exist. Dative objects are only expressed using an
applicative type structure. Adopting the work of Synder (2001), this suggests Wolof does
not have Principle R in the interpretational component of the grammar. I therefore
conclude that the lack of resultative and to-dative small clauses excludes a Low
Applicative or small clause analysis for dative applicatives. The applicative head
proposed by Georgala (2012) for raising applicatives, ApplE, combines with the predicate
via Event Identification and therefore does not need Principle R.
3.4. Conclusion
In this chapter, I have shown that both benefactive and dative applicatives are
symmetrical and they share object and c-command properties. Despite these similarities,
dative applicatives do not allow unspecified object deletion. They require an
detransitivizing morpheme, -e, for the one of the objects to be suppressed while
benefactive applicatives allow unspecified object deletion of the theme object. Due to this
difference, I conclude that benefactive and dative applicatives do not share the same
structure; they result from different derivations.
!127
I argue that benefactive applicatives have a Thematic Applicative structure and datives
have a Raising Applicative structure (Georgala 2012). Benefactive objects are selected by
the ApplT head while dative objects are selected by the lexical verb and raise to the
ApplE projection. As we will see in the next chapter, instrumental and locative
applicatives will require a third type of applicative structure.
!128
Chapter 4
4. OBLIQUE APPLICATIVES
This chapter deals with the two remaining types of applicatives found in Wolof:
instrumental and locative. These applicatives exhibit variation in form and behavior to a
much higher degree than their benefactive and dative counterparts. Because of this
variation I have included data from all types of instrumental and locative constructions,
applicative and non-applicative. The main focus, however, will be on what I call the in
situ applicative since this is the closest to the benefactive and dative applicatives and will
show the most about VP structure. In section 4.4, I argue instrumental and locative in situ
applicatives involve a third type of Appl head, Oblique Appl (abbreviated as ApplO).
Oblique Appl is similar to Thematic Appl in that it relates an individual to an event but
unlike ApplT, it merges downward with the VP.
4.1. Oblique objects in Wolof
To begin the discussion of Wolof instrumental and locative applicatives, let’s consider the
different types of constructions that can express instrumental and locative objects. First,
in in situ applicatives the applied object appears in situ within the VP. The form of the
applicative suffix in instrumental and locative sentences is -e.1 The instrumental object,
big ‘pen’, appears to the right of the theme bataaxal bi ‘the letter’.
(1)
Moustapha dafa-y
[bind-e
bataaxal bi
big]VP.
Moustapha 3SG.AFF-IMP write-APPL letter
DEF1 pen
‘Moustapha is writing the letter with a pen.’
In the second type of oblique constructions, the applied object is focused and
appears in a sentence initial position outside of the VP in conjunction with object focus
morphology on the subject marker, laa in the following example.
1
Not to be confused with the homophonous detransitivizing morpheme, -e, discussed in chapter 3.
!129
(2)
Big
laa
[bind-e
pen
1SG.OFOC
write-APPL
It’s a pen with which I wrote the letter.’
bataaxal
letter
bi]VP.
DEF1
Third, instrumental and locative objects can also be expressed without using an
applicative verb. In such cases, the instrument or locative is contained within a
prepositional phrase similar to English, such as ak big ‘with a pen’ below.
(3)
Dama-y
[bind
bataaxal bi
1SG.AFF-IMP write
letter
DEF1
‘I am writing the letter with a pen.’
[ak
with
big]PP]VP.
pen
A fourth way of expressing instrumental and locative objects combines the in-situ
applicative and the prepositional variant. I call these hybrid applicatives. Hybrid
applicatives have the applicative suffix -e associated with the verb but the “applied”
object is still contained in a prepositional phrase.
(4)
Dama-y
[bind-e
bataaxal
1SG.AFF-IMP write-APPL letter
‘I am writing the letter with a pen.’
bi
DEF1
[ak
with
big]PP]VP.
pen
I will discuss in-situ applicative first, followed by hybrid applicatives because
these tell the most about applicative VP structure. I outline the prepositional variant as
well since it will be contrasted with in situ and hybrid applicatives throughout the
chapter. I discuss the fronted applicatives last because they do not show anything about
VP structure but are frequently used in spoken and written Wolof.
The object properties of instrumental and locative applicatives are given in
section 4.2. Section 4.3. presents c-command facts about instrumental and locative
applicatives. The analysis of the instrumental and locative applicatives is presented in 4.4.
!130
4.1.1. In situ applicatives
In-situ applicatives contain applicative morphology and allow two post verbal direct
objects like benefactive and dative applicatives. Their productivity is limited and
sentences of this type are deemed marginal by native speakers. However, they aren’t
judged to be ungrammatical and speakers do have intuitions about them so therefore I
include them, setting aside the question as to why they are less natural than benefactive
and dative applicatives. In situ applicatives in Wolof are also mentioned in the literature
by Dione (2013) for both instrumental and locative applicatives.
Instrumental in situ applicative
(5)
Dama-y
bind-e
bataaxal
1SG.AFF-IMP write-APPL
letter
‘I am writing the letter with a pen.’
bi
DEF1
Locative in situ applicative
(6)
Ma ngi-y
jàng-e
téere
1SG 1SG.IMP
read-APPL
book
‘I am reading the book at my house.’
bi
DEF1
big.
pen
sama
1SG.POSS
kër.
house
Unlike benefactive and dative applicatives, applied instrumental and locative
objects can only appear following the theme object. When the instrumental or locative
object is adjacent to the verb as in (7),where big ‘pen’ precedes bataaxal bi ‘the letter’,
the sentence is categorically rejected (compare to 5).
(7)
*Damay
bind-e
big
bataaxal
1SG.AFF
write-APPL pen
letter
‘I am writing the letter with a pen.’
bi.
DEF1
Dione (2013) also confirms that instrumental and locative objects must follow the theme.
Similarly, the applied locative object sama kër ‘my house’ cannot precede the
theme object téere bi ‘the book’, as seen when comparing (8) and (6) above.
!131
(8)
*Ma ngi-y
jàng-e
sama
1SG
1SG-IMP read-APPL 1SG.POSS
‘I am reading at my house the book.’
kër
house
téere
book
bi.
DEF1
This rigid word order is not seen in benefactive or dative applicatives, as we saw in
chapter 3, section 3.1.1.
In situ applied objects are compatible with intransitive predicates. Again, the
applicative suffix -e is associated with the verb. The applied object, yet ‘cane’ in (9) and
kër ga ‘the house’ in (10), appears without a preposition.
(9)
Dama-y
dox-e
yet.
1SG.AFF-IMP walk-APPL cane
‘I am walking with a cane.’
(10)
Dama-y
togg-e
kër
1SG.AFF-IMP cook-APPL house
‘I am cooking at the house.’
ga.
DEF1
I note here that there is one example of a locative verb that is obligatorily
ditransitive, the verb gunge ‘to accompany’. Unlike the other in situ applicatives we’ve
seen, there is no applicative morphology on gunge. Despite these differences, the word
order is fixed, theme - locative, like other in situ locative applicatives.
(11)
Dama-y
gunge
Issa
1SG.AFF-IMP accompany
Issa
‘I am accompanying Isa to Dakar.’
Dakar.
Dakar
(12)
*?Dama-y
gunge
Dakar
1SG.AFF-IMP accompany
Dakar
‘I am accompanying Isa to Dakar.’
Issa.
Issa
!132
I assume that sentences with gunge are examples of valency-preserving applicatives
(Creissels 2004) because the locative object appears as a direct object of the verb rather
than inside a prepositional phrase, meaning that the verb has two direct objects. In fact,
the locative object cannot appear in a prepositional phrase.
(13)
*Dama-y
gunge
Issa ci
1SG.AFF-IMP accompany Issa to
‘I am accompanying Isa to Dakar.’
Dakar.
Dakar
Even though there is not a valency increase present with gunge, I assume it involves
applicative structure since the locative object is brought forward by being made an object
of the verb.2
Wolof is not the only language where instrumental and locative applicative show
inverse word order of benefactive or dative applicatives. Locative applicatives in
Kiswahili (14) and instrumental applicatives in Kinyarwanda (15) also have theme locative word order like Wolof.
Kiswahili Locative
(14) a.
wa-teja
wa-li-l-i-a
ch-akula
2-customer 2-PST-eat-APPL-FV 7-food
‘The customers ate food in the office.’
ofisi-ni.
9.office-LOC
(Ngonyani 1998:83)
Kiswahili Locative
b.
*wa-teja
wa-li-l-i-a
ofisi-ni
2-customer 2-PST-eat-APPL-FV 9.office-LOC
‘The customers ate food in the office.’
2
ch-akula
7-food
(Ngonyani 1998:81)
See chapter 2, section 2.1. for more on treating lexically ditransitive verbs as applicatives.
!133
Kinyarwanda Instrumental
(15) a.
Umukoôbwa a-ra-andik-iish-a
íbárúwa íkárámu.
girl
she-PRES-write-INSTR-ASP letter
pen
‘The girl is writing a letter with a pen.’
Kinyarwanda Instrumental
b.
Umubooyi a-ra-kat-iish-a
inyama
cook
she-PRES-cut-INSTR-ASP meat
‘The cook is cutting meat with a knife.’
ícyúuma.
knife
(Kimenyi 1980:32)
The data from Wolof match Kiswahili and Kinyarwanda and show the applied
object, be it instrumental or locative, cannot appear adjacent to the verb. This is the
opposite pattern from what was seen with benefactive and dative applicatives in chapter 3
and will be taken into consideration in the analysis proposed in section 4.4. Now, I turn to
hybrid applicatives.
4.1.2. Hybrid Applicatives
The hybrid construction involves the applicative suffix -e on the verb used in conjunction
with a prepositional phrase. Dione (2013) and Creissels (2004) calls these constructions
non-canonical applicatives because they contain an object, in the form of a PP, that the
same non-derived verbs can license. This construction is accepted by all consultants
unlike the in situ applicative examples.3
Instrumental hybrid applicative
(16) Damay
bind-e
bataaxal bi
1SG.AFF write-APPL letter
DEF1
‘I am writing the letter with a pen.’
ak
big.
with pen
At this point, it is unclear if there is a semantic or pragmatic difference between the hybrid applicatives
and in situ applicatives. Schwartz (1975) and Dunigan (1994) suggest that, at least in locative
constructions, the suffix -e signals that the prepositional phrase is an argument of the verb.
3
!134
Locative hybrid applicative
(17) Sindax bi
laa
rey-e
sa
lizard
DEF1 1SG.OFOC kill-APPL at
‘It was a lizard that I killed under the tree.’
garab
tree
ba.
DEF.DIST
Hybrid applicatives also have theme - instrument/locative default word order like
we saw above for in situ applicatives. It is unclear if the instrumental PP is allowed to
intervene between the verb and the theme. Grammaticality judgements of my data are
simply too varied to draw a conclusion.
4.1.3. Prepositional instruments and locatives
I now discuss the simple prepositional variant, which involves a prepositional phrase and
no applicative morphology on the verb. The verb takes two complements, a direct object
and a prepositional phrase. This structure is analogous to prepositional complement
sentences in English. In (18) the theme object, bataaxal bi ‘the letter’, is the direct object
of the verb and is followed by a prepositional phrase which contains the instrument, ak
big ‘with a pen’. A prepositional locative, ci kër ga ‘in/at the house’, is shown in (19).
Instrument
(18) Dama-y
bind
bataaxal bi
1SG.AFF-IMP write
letter
DEF1
‘I am writing the letter with a pen.’
Locative
(19) Dama-y
nafar
téere
[ci
1SG.AFF-IMP read
book
in/at
‘I am reading a book at the house.’
[ak
with
kër
house
big]PP.
pen
ga]PP.
DEF1
Instrumental and locative prepositional phrases are also compatible with
intransitive predicates.
!135
Instrument
(20) Dama-y
dox
[ak
1SG.AFF-IMP walk
with
‘I am walking with a cane.’
yet]PP.
cane
Locative
(21) Dama-y
togg
[ci
1SG.AFF-IMP cook
in/at
‘I am cooking at the house.’
kër
house
ga]PP.
DEF1
Prepositional constructions do allow the instrumental PP (22) and locative PP (23) to
intercede between the verb and the theme.
Instrument
(22) Dama-y
bind
[ak
big]PP bataaxal
1SG.AFF-IMP
write
with
pen
letter
‘I am writing the letter with a pen.’
Locative
(23) Jél
naa
[sa
ja ba]PP
take
1SG
at
market
‘I took photos at the market.’
ay
DEF.PL
bi.
DEF1
natal.
photo
This behaviour is different than in situ applicatives, where the instrument or
locative cannot precede the theme.
4.1.4. Fronted Oblique Objects
As previously mentioned, applied instrumental and locative objects often appear at the
beginning of the sentence in a focus position. Applicative morphology on the verb is
required for the sentence to be grammatical.
!136
(24)
Big
laa
bind-*(e)
pen
1SG.OFOC
write-APPL
It’s a pen with which I wrote the letter.’
bataaxal
letter
bi.
DEF1
In (24), the instrumental applied object, big ‘a pen’, is in the fronted position, and object
focus morphology is used, laa. The applicative suffix -e is attached to the verb, bind ‘to
write’ and the theme object, bataaxal bi, is in its canonical position following the verb.
The sentence is ungrammatical without the suffix -e. Locative objects follow the same
pattern as illustrated in (25) and (26).
(25)
(26)
Sama
kër
laa-y
jàng-e
téere
1SG.POSS house 1SG.OFOC-IMP read-APPL book
‘It is at my house where I am reading the book.’
bi.
DEF1
Kër
Boris
laa
togg-e
ceebu-jën.
house Boris
1SG.OFOC cook-APPL rice-fish
‘It was at Boris’ house where I cooked ceebu-jën (a rice and fish dish).’
Fronted instrumental and locative objects are also compatible with intransitive
predicates. Instrumental examples are shown in (27)-(28) and (29) shows an analogous
locative example.
(27)
Woto
laa
car
1SG.OFOC
‘I went by CAR.’
dem-e.
go-APPL
(28)
Yet
laa-y
cane
1SG.OFOC-IMP
I walk with a CANE.’
dox-e.
walk-APPL
!137
(29)
Kër
Boris
laa
togg-e.
house
Boris
1SG.OFOC cook-APPL
‘It was at Boris’ house where I cooked.’
Before continuing, it is worth noting that locative objects can be associated with
some intransitive verbs without a preposition or applicative morphology. For example,
one can say “I am going to Dakar” without the use of a prepositional phrase in Wolof.
(30)
Dama-y
dem
Ndakaaru.
1SG.AFF-IMP go
Dakar
‘I am going to Dakar.’
With this verb, the locative object can also be fronted without applicative morphology
(31).
(31)
Ndakaaru laa
dem.
Dakar
1SG.OFOC go
lit: ‘It is to Dakar that I am going.’
In fact, using a preposition, like ci, or the applicative suffix, -e, with the verb dem ‘to go’
gives rise to an unacceptable sentence.
(32)
a.
*Dama-y
dem ci
1SG.AFF-IMP go
to
‘I am going to Dakar.’
Ndakaaru.
Dakar
b.
*Dama-y
dem-e
Ndakaaru.
1SG.AFF-IMP go-APPL Dakar
‘I am going to Dakar.’
These examples are similar to the gunge example given in (11) through (13). The
difference is that the locative object associated with gunge is obligatory while the
locatives in (30) and (31) are optional. This pattern of locative objects surfacing without a
!138
preposition is found with certain deictic verbs in Wolof. This class of verbs deserves
further study but for present purposes, despite the lack of morphology, I assume they are
applicatives but leave aside their relation to morphologically derived and ditransitive
locative applicatives.
4.2. Object Properties
This section looks at the object properties in in situ and hybrid applicatives and in
prepositional phrases. We’ve already seen that word order in instrumental and locative
applicatives is fixed. This section shows instrumental and locative applicatives show
other symmetrical applicative behaviour with a twist. Both object can be pronominalized
with clitics but only the applied object can be A-bar extracted.
4.2.1. Pronominalization
Recall from chapter 3 that clitic pronouns in Wolof can replace applied objects like they
replace direct objects in transitive sentences. Instrumental objects can only be
pronominalized with a clitic in in situ applicatives. Locative pronouns have a different
form than instrumental objects in both the clitic and strong pronoun paradigms and will
be shown to be insensitive to objecthood. Instrumentals will therefore be discussed first,
followed by the locative pronominalization data.
Instrumental objects are replaced with clitic pronouns in in situ applicatives,
illustrated in (33) and (34), just like benefactive and dative applied objects (see chapter
3, section 3.1.2).
(33)
Dama
ko-y
bind-e
1SG.AFF 3SG.OBJ-IMP write-APPL
‘I am writing the letter with it.’
bataaxal
letter
bi.
DEF1
!139
(34)
Dama
ko-y
1SG.AFF
3SG.OBJ-IMP
‘I am walking with it.’
dox-e.
walk-APPL
Strong pronouns are used when the instrument is contained in a prepositional
phrase or is fronted, as illustrated in (35). The instrument has been pronominalized with
the strong pronoun moom ‘he/she/it’. The clitic pronoun ko cannot be used in either case.
(35)
a.
Dama-y
dox
ak
1SG.AFF-IMP walk with
‘I am walking with it.’
moom/*ko.
3SG.OBJ
b.
Moom/*ko laa-y
dox-e.
3SG.OBJ 1SG.OFOC-IMP walk
‘I am walking with IT.’ (lit: It is it with which I am walking.)
Theme objects are also replaced by the clitic pronoun ko ‘him/her/it’ in in situ
applicatives (36a) as well as in prepositional complement sentences (36b).
(36)
a.
Dama
ko-y
bind-e
big.
1SG.AFF 3SG.OBJ-IMP write-APPL pen
‘I am writing it with a pen.’
b.
Dama
ko-y
bind
1SG.AFF 3SG.OBJ-IMP write
‘I am writing it with a pen.’
ak
with
big.
pen
When both objects are pronominalized with the 3SG clitic pronoun ko, as in (37a),
speakers find the sentence odd. Speakers preferred the version with only one pronoun, ko,
replacing the applied instrumental object and the theme object left unspecified (37b).
!140
(37)
a.
?Dama
ko
ko-y
bind-e.
1SG.AFF 3SG.OBJ 3SG.OBJ-IMP write-APPL
I am writing it with it.’
b.
Dama
ko-y
bind-e.
1SG.AFF 3SG.OBJ-IMP
write-APPL
‘I am writing (something) with it.’
Given that two ko’s are permitted in benefactive and dative applicatives, it is not likely
that two ko’s are blocked in instrumental applicatives for purely phonological reasons.
(38)
Bind-al
na
ko
write-APPL 3SG
3SG.OBJ
‘He/she wrote it for him/her.’
ko.
3SG.OBJ
These data indicate that instrumental applicatives are symmetrical because like
dative and benefactive applicatives, either object can be pronominalized with the clitic
pronoun. Unlike benefactive and dative applicatives, only one object can be
pronominalized at a time. Looking at extraction in 4.2.2., we will see that these
applicatives exhibit a mix of properties, both symmetrical and asymmetrical.
Turning now to locative constructions, the pronominalization patterns for locative
objects are different from the other types of objects. Recall that benefactive, dative, and
instrumental applied objects are all pronominalized using the clitic pronouns listed in
Table 1 from chapter 3, repeated here for convenience.
Table 1
1SG
2SG
3SG
1PL
2PL
3PL
Clitic
ma
la
ko
nu
leen
leen
Strong
man
yow
moom
nun
yeen
ñoom
!141
Locative objects cannot be replaced by pronouns in this paradigm, neither clitic nor
strong. They are replaced by either the locative clitic pronoun fi/fa ‘here/there’ or the
strong locative pronoun foofu ‘there’. Unlike the pronouns in Table 1, the choice of a
clitic or strong locative pronoun isn’t determined by object status of the object. Unlike
clitic and strong pronouns in Table 1, fi/fa and foofu are not in complementary
distribution with each other. The speaker has the choice as to which form to employ. To
see this, I have created question-answer pairs that target locative expressions. In (39a) the
locative is an object, not contained in a PP, while in (40a) the locative, butig ba ‘the store’
is inside a PP. The (b) sentences show that in both cases, answers using either fa or foofu
are grammatical.
(39)
(40)
a.
Gis nga ma
dëkk
ba?
see 2SG 1SG.OBJ village DEF.DIST1
‘Did you see me at the village?’
b.
Waaw, gis naa
la
fa/foofu.
yes
see 1SG 2SG.OBJ there
‘Yes, I saw you there.’
a.
Jënd nga
suukër ak meew ca
butig
sell 2SG
sugar and milk PREP1 store
‘Do you sell sugar and milk at the store?’
b.
Waaw, jënd
naa
yes
sell
1SG
‘Yes, I see them there.’
ba?
DEF.DIST
leen
fa/foofu.
3PL.OBJ there
Thus, pronominalization is not a very useful test for the objecthood of locative objects
because the choice of pronoun does not reveal if the locative is an object of the verb or
the object of a preposition.
!142
Looking at the data from pronominalization tests, the applied and theme objects
can be pronominalized using clitic pronouns in instrumental in situ applicatives. In hybrid
applicatives and prepositional complement sentences the theme can be replaced with a
clitic pronoun but the applied must be replaced with a strong pronoun. This means that in
in situ applicatives, both objects behave as direct objects of the verb while in hybrid
applicatives and prepositional complements only the theme does. This confirms my
decision to focus the analysis on in situ applicatives and leave the structure of hybrid
applicatives for a later date.
4.2.2. Extraction in oblique applicatives
I now look at extraction of objects in instrumental and locative applicatives and we will
see that only the applied object can be extracted. In in situ applicatives, the applied
object, either locative or instrumental, can be questioned or focused, but the theme object
cannot.
Locative questioned
(41) a.
Fan
nga
togg-e
ceebu-jën?
(>Foo)
where 2SG
cook-APPL ceebu-jën
‘Where did you cook ceebu-jën?’
Theme questioned
b.
*Lan nga togg-e
kër
Boris?
what 2SG cook-APPL house Boris
‘What did you cook at Boris’ house?’
‘With what did you cook at Boris’ house?’
Locative focused
(42) a.
Kër Boris laa
togg-e
ceebu-jën
house Boris 1SG.OFOC cook-APPL rice.and.fish
‘It was at Boris’ house where I cooked rice and fish.’
!143
Theme focused
b.
*Ceebu-jën laa
togg-e
kër
Boris
rice.and.fish 1SG.OFOC cook-APPL house Boris
‘It was rice and fish that I cooked at Boris’ house.’
Instrument questioned
(43) a.
Lan
nga
rey-e
mbóot
what
2SG
kill-APPL roach
With what did you kill the cockroach?’
mi?
DEF
Theme questioned
b.
*Lan
nga
rey-e
what
2SG
kill-APPL
What did you kill with the shoe?’
bi?
DEF
dàll
shoe
Instrument focused
(44)
a.
Paakaa bi
laa
dagg-e
yàpp
knife
DEF1 1SG.OFOC cut-APPL meat
It was the knife with which I cut the meat.’
Theme focused
b.
*Yàpp yi
laa
dagg-e
meat
DEF.PL 1SG.OFOC
cut-APPL
‘It was the meat that I cut with the knife.’
yi.
DEF.PL
paakaa
knife
bi.
DEF1
The (a) sentences show extraction of the locative and instrumental objects. Each of these
sentences is grammatical so we know the applied object can be extracted in Wolof. The
(b) sentences show extraction of the theme object. None of these sentences are
grammatical leading to the conclusion that the theme cannot be extracted. This is unlike
benefactive and dative applicatives which allow extraction of either object (see chapter 3,
section 3.1.3).
!144
In prepositional complement sentences, the results are different. The theme can be
questioned or focused but not the instrument or locative object. Prepositional locatives
are considered first in examples (45) and (46). Prepositional instruments are shown in
examples (47) and (48).
Theme questioned - locative
(45) a.
Lan
nga togg
ci
kër
Boris?
what 2SG cook
at
house Boris
‘What did you cook at Boris’ house?’
Locative questioned
b.
*Fan nga
togg
ceebu-jën?
what 2SG
cook
rice.and.fish
‘Where did you cook rice and fish?’
Theme focused
(46) a.
Ceebu-jën laa
lekk sa ja ba.
ceebu-jën 1SG.OFO cook at market
C
It was ceebu-jën that I ate at the market.’
Locative focused
b.
*Sa ja ba
laa
lekk ceebu-jën.
at
market 1SG.OFO eat
rice.and.fish
C
‘It was at the market where I ate rice and fish.’
Prepositional instruments show the same pattern as prepositional locatives. In the
prepositional complement sentence, only the theme, yàpp yi ‘the meat’ can be extracted
for questioning or focus.
Theme focused
(47) a.
Yàpp yi
naa dagg
ak paakaa bi?
meat DEF.PL 1SG cut-APPL with knife
DEF1
‘It was the meat that did you cut with the knife?’
!145
Instrument focused
b.
*Paakaa
bi
naa dagg
knife
DEF1 1SG
cut
‘With what did you cut the meat?’
yàpp
meat
Theme questioned
(48) a.
Lan
nga
bind
what
2SG
write
‘What did you write with the pen?’
ak
with
yi?
DEF.PL1
big?
pen
Instrument questioned
b.
*Lan
nga
bind
bataaxal
what
2SG
write
letter
‘With what did you write the letter?’
bi?
DEF1
Sentences with an extracted instrument or locative prepositional phrase, like
(48b), cannot be saved by realizing the preposition either fronted or stranded after the
theme.
Extracted full PP
(49) a.
*Ak lan
nga dagg
with what
2SG cut
‘With what did you cut the meat?’
yàpp
meat
Stranded preposition
b.
*Lan
nga dagg
yàpp bi
what
2SG cut
meat DEF1
‘With what did you cut the meat with?’
bi?
DEF1
ak?
with
Their ungrammaticality means the the prepositional phrase cannot be extracted. Next I
look at extraction of instrumental and locative PPs in hybrid applicatives.
!146
Hybrid applicatives are in between prepositional complements and in situ
applicatives. They allow the theme to be extracted like we saw above for prepositional
complements. They also allow a full locative PP to be focused as seen in (50), unlike
prepositional complement sentences.
Theme focused
(50) a.
Xale bi
laa
dóor-e
sa ja ba.
child DEF1 1SG.OFOC hit-APPL at market
‘It was the child that I hit at the market.’
Locative PP focused
b.
Sa géej
ba
laa
napp-e
ay
at
beach DEF1 1SG.OFOC catch-APPL DEF.PL
It was at the beach where I caught fish.’
jën.
fish
Themes can also be extracted in instrumental prepositional complements, as seen with
yàpp yi ‘the meat’ (51a). Extracting an instrumental PP, like ak yet bi ‘with the cane’,
even in the hybrid applicative, is questionable but not categorically rejected by speakers.
Theme focused
(51) a.
Yàpp yi
laa
dagg-e
ak
paakaa
meat DEF.PL 1SG.OFOC cut-APPL with knife
‘It was the meat that I cut with the knife.’
Instrumental PP focused
b.
?Ak yet
bi
laa
dóor-e xale
with cane DEF1 1SG.OFOC hit-APPL child
1 with which I hit the child.’
‘It was the cane
bi.
DEF1
bi.
DEF1
Only applied objects can be extracted in in situ applicatives. Both the theme and
applied PP object can be extracted in hybrid applicatives although some speakers
disagree with extracted instruments. Finally, only theme objects can be extracted in
!147
prepositional complement sentences. The relevant patterns of extraction are illustrated in
Table 2.
Table 2. Extraction of objects
In-situ
Hybrid
Prepositional
Theme
(in locatives and instrumentals)
N
Y
Y
Locative
Y
Y
N
Instrument
Y
Y*
N
*Allowed but marginal
Summing up, the tests for the object properties of instrumental and locative
applicative structures reveal mixed results for theme and applied objects. Instrumental
and locative objects in in situ applicatives overwhelmingly pattern with direct objects.
They are pronominalized with clitic pronouns like direct objects. They can be questioned
and focused like direct object of a transitive verbs. On the other hand, they are not
allowed to intervene between the theme object and the verb, which does not match the
behaviour of direct objects in transitive sentences. Turning now to the theme, it is
pronominalized with a clitic pronoun just like direct objects of transitive verbs and
appears adjacent to the verb, but it cannot be extracted, unlike objects of transitive
sentences. So in instrumental and locative applicatives, the applied object looks like a
direct object except verb adjacency and the theme looks like a direct object except for
extraction. The analysis presented in section 4.4 will account for these facts. Although I
classify instrumental and locative in situ applicatives as symmetrical, I acknowledge that
the properties considered are not uniform in Wolof, raising again the question of the
usefulness of binary opposition between “symmetrical” and “asymmetrical”.
4.3. C-command
Recall from chapters 2 and 3 that certain tests can be used to determine the c-command
relationship between two objects in an applicative structure. In this section, I apply
!148
quantifier binding and weak cross-over tests to instrumental and locative applicatives to
identify the c-command relationships in in situ and hybrid applicatives, and in
prepositional complement constructions. Data from the applicatives show theme objects
asymmetrically c-command applied objects. The c-command results for fronted
applicatives do not offer insight to the structure of the VP so they are not included.
4.3.1. Quantifier Binding
In what follows, I use comparative judgements of acceptability to test for quantifier
binding between the objects. To start with, an applied instrumental object cannot bind a
theme object. That is to say, a universal quantifier associated with the applied instrument
cannot bind an anaphor in the theme. In (52a) the theme object appears first since applied
instrumental and locative objects follow themes, as we saw earlier. The quantified
instrumental object, paaka bu nekk ‘every knife’ cannot bind the anaphor of the theme
object, borom-om ‘its owner’. If the anaphor in theme position -om ‘its’ is not interpreted
as coreferential with the quantified instrument, then the sentence is grammatical.
(52)
a.
Dagg-e
naa borom-om*i/j
cut-APPL
1SG owner-3SG.POSS
I cut its*i/j owner with every knifei.’
paaka
knife
bu
nekki.
CREL each
Binding cannot be achieved by placing the instrument before the theme, as seen in
(53). As already shown in section 4.1.1. above, object scrambling in instrumental and
locative applicatives is not permitted; instrumental and locative applied objects cannot
precede theme objects. The sentence in (53) is therefore ungrammatical, whether or not
the pronoun is interpreted as bound.
(53)
*Dagg-e
naa
paaka bu
cut-APPL
1SG
knife CREL
‘I cut with every knifei itsi owner.’
nekki
each
borom-omi.
owner-3SG.POSS
!149
Quantifier binding thus shows that in Wolof an applied instrumental object does
not c-command the theme object. The same results obtain for locative in situ applicatives.
Quantified locative objects cannot bind anaphors in the theme object.
(54)
a.
Gunge
naa dawalkat-am*i/j
oto
accompany 1SG driver-3SG.POSS car
I accompanied itsi driver to every cari.”
bu
nekki.
CREL exist
Now I look at the possibility of the theme c-commanding the applied object. A
quantified theme object can bind an anaphor in the applied object. In the following
example, the quantified theme, xale bu nekk ‘every child’, binds the anaphor in the
applied instrumental object paakaam ‘his knife’.
(55)
Dagg-e
naa xale bu
nekki
cut-APPL 1SG child CREL exist
‘I cut each childi with hisi knife.
paaka-ami.
knife-3SG.POSS
Locative applicatives show the same pattern as instrumentals. Quantified themes can bind
the anaphor in the locative object (56a).
(56)
a.
Gunge
naa
góor [gu] nekki
accompany 1SG man CREL exist
“I accompanied every mani to hisi house.”
b.
*Gunge
naa kër-ami
góor
accompany 1SG house-3SG.POSS man
“I accompanied every mani to hisi house.”
kër-ami.
house-3SG.POSS
[gu]
CREL
nekki
exist
These data are similar to Kinyarwanda instrumental applicatives. McGinnis (2005)
provides examples showing a theme object c-commands the instrumental object in
Kinyarwanda, much like the theme object in Wolof. The quantified theme buri muryango
‘each door’ binds the anaphor in the instrumental NP úrufunguz rwáwo ‘its key’. In (57b)
the sentence is ungrammatical under a coreferential reading as indicated by the indices.
!150
A. Rutayoberana (p.c.); cited in McGinnis (2005:195)
(57) a.
N-a-fúngul-ish-ije
buri muryangoi
I-PST-open-INST-ASP each door
‘I opened each doori with itsi key.
b.
úrufunguzo rwáwoi.
key
its
N-a-fúngul-ish-ije
umuryano wáyoi
I-PST-open-INST-ASP door
its
‘I opened itsi door with each keyj/*i.’
buri
each
rufunguzo*i/j.
key
This confirms the hypothesis that the theme c-commands the instrument as in Wolof.
Given these data, I conclude that the theme asymmetrically c-commands the
applied object, be it instrumental or locative.
4.3.2. Weak Cross-Over
Recall from section 4.2.2. that theme objects cannot be questioned in in situ applicative
constructions. This is not entirely accurate. Speakers consider questioning non-D-linked
themes ungrammatical (58a). If the theme is D-linked, the question improves
dramatically (b).
Non D-linked theme
(58) a.
*Lan nga dagg-e
t paakaa
what
2SG cut-APPL
knife
‘What did you cut with the knife?’
bi?
DEF1
D-linked theme
b.
?Ban
yàpp nga dagg-e
t
paaka
which meat 2SG cut-APPL
knife
‘Which meat did you cut with the knife?
bi?
DEF1
!151
Non D-linked theme - in situ locative applicative
(59) a.
*Lan nga togg-e
t
kër
what
2SG cook-APPL
house
‘What did you cook at Boris’ house?’
Boris?
Boris
D-linked theme
(b)
?Ban
yàpp
nga
togg-e
which
meat
2SG
cut-APPL
t
kër
Boris.
knife
DEF1
‘Which meat did you cook at Boris’ house?’
Now that we have a way to question theme objects in applicatives in Wolof, we
can use Weak Cross-Over to investigate c-command. No weak cross-over violation arises
when the theme object is questioned and fronted. The theme, ban góor ‘which man’ is
coreferential with the instrument, paakaam ‘knife’ in (60a). On the other hand,
questioning the applied object leads to a weak cross-over violation. The applied object,
yetu kan ‘whose cane’ cannot be referential with the theme, yaayam ‘his/her mother’ in
(60b).
(60)
a.
Ban
góori nga d-oon
dagg-e t paaka-ami?
which man 2SG IMP-PST cut-APPL knife-3SG.POSS
“Which mani were you cutting with hisi knife?”
b.
Yet-u
kani nga dóor-e
yaay-am*i/j
cane-CS who 2SG hit-APPL mother-3SG.POSS
‘(With) whose cane did you hit his mother?
t?
Similar facts obtain for locative applicatives. The sentence in (61) is deemed
grammatical but marginal when the wh-phrase kan ‘who’ is coreferential with the
anaphor in the locative object këram ‘his house’. When the applied object is questioned,
however, it cannot be coreferential with the anaphor in the theme object indicating a
weak cross-over violation. If the wh-phrase and anaphor are not coreferential, the
sentence is grammatical.
!152
(61)
a.
?Xariti-u
kanj nga gunge
t kër-amj?
friend-CS
who 2SG accompany house-3SG.POSS
“Whosei friend did you accompany to hisi? house?”
b.
Kër-u
kani nga gunge
xarit-am*i/j t ?
house-POSS who 2SG accompany friend-3SG.POSS
“To whosei house did you accompany his*i/j friend.”
I note here that I was unable to test for weak cross-over with locative applicatives
with overt morphology, like daaj-e ‘nail’. The sentences, such as (62) were all considered
ungrammatical. Even in English, the equivalent sentences are not acceptable, perhaps
because of the inanimacy of the theme object.
(62)
*Toggukay-u kani nga daaj-e
béréb-u liggeykay-ami?
chair-CS
who 2SG nail-APPL site
work-3SG.POSS
‘*Whose chairi did you nail (together) at hisi workshop?
(63)
*Which fishi did you catch in itsi ocean?
These sentences were simply too odd to be acceptable so testing with such predicates did
not provide any information about VP structure.
For comparison, let’s look at weak cross-over in the prepositional complement
construction. Like in situ applicatives, the theme is allowed to be questioned without
incurring a weak cross-over violation (64a). The oblique object cannot be questioned
because PPs cannot be extracted and the sentence in (b) does not reveal anything about
weak cross-over or VP structure.
(64)
a.
Ban
xalei nga dóor t ak
yet-ami?
which child 2SG hit
with cane-3SG.POSS
‘Which childi did you hit with hisi cane?’
!153
b.
*Ak yet-u
kani nga dóor doom-ami
t?
with cane-CS who 2SG hit
child-3SG.POSS
‘With whosei cane did you hit his/her*i/j child?
Instrumental and locative applicatives share the same c-command pattern as the
prepositional complement construction, where the theme consistently c-commands the
instrument or locative. This pattern is different than what was seen with benefactive and
dative applicatives.
4.3.3. Conclusion
C-command patterns for instrumental and locative applicatives do not match the ccommand patterns of benefactive or dative applicatives. There are two differences
between dative and benefactive applicatives on the one hand and instrumental and
locative applicatives on the other vis-à-vis c-command. First, benefactive and dative
applicatives show variable c-command depending on the linear order of the objects for
quantifier binding and reflexivization. The first object c-commands the second regardless
of theta role. However, variable c-command is not seen in instrumental and locative
applicatives with quantifier binding since object scrambling is not possible. The second
difference is revealed through weak cross-over. In benefactive and dative applicatives, the
applied object asymmetrically c-commands the theme. The opposite is true of
instrumental and locative applicatives; the theme asymmetrically c-commands the
applied object.
4.4. Analysis
In chapter 3, it was established that a single Appl structure for all applicative types is not
tenable. Benefactive and dative applicatives motivated two Appls: thematic for
benefactive applicatives and raising for dative applicatives. The instrumental and locative
data presented thus far in chapter 4 also support a multiple Appl approach to applicatives
but are not adequately accounted for using only Thematic and Raising Appls. I argue in
!154
this section that instrumental and locative applicatives involve a third type of Appl
structure. This Appl, which I will call Oblique Appl (ApplO), selects an an event and an
instrumental or locative object, much like Thematic Appl but with one crucial difference.
Oblique Appl merges with the VP in a downward fashion, following the work of
McGinnis (2005) (see also Philips (2003) and Richards (2002) for more on downward
merge).
Recall from chapter 2 (section 2.4) that McGinnis argues Merge can occur upward
or downward and that two elements can merge without necessarily becoming sisters. This
is the case with ApplO. The figure in (63a) shows the representation of Appl that has
merged upward with the VP. The figure in (63b) shows the representation of Appl which
has merged downward with the VP (AO stands for applied object).
(65)
a.
b.
In both examples, the DO receives the theme role because it merges with V. This
relationship is obvious in (63a) but more opaque in (63b). The opacity is because the
subsequent merge of the ApplO head leads to a reanalysis of the existing structure and the
DO becomes the specifier of ApplO. The relationship created from the original V + DO
merge is not broken by the reanalysis, however, and the DO retains its theme role. Having
merged outside the event at the level of VP, ApplO introduces the instrumental or locative
object as a VP-external object, like benefactive objects. The fact that the applied object is
merged above the VP and is never a sister to a V projection is important and will be
discussed further in 4.4.3.
!155
Let’s see how this works in Wolof. I start with an intransitive verb (64) for clarity
and then extend the analysis to transitive verbs. Since the verb doesn’t not select an
object, it merges directly with ApplO. Then the instrument big ‘pen’ merges downward
with ApplO.
(66)
a.
(67)
a.
Damay bind-e
big.
1SG.AFF write-APPL pen
‘I am writing with a pen.’
b.
Now let’s look at a transitive example. In (68), there are two objects, the theme
bataaxal bi ‘the letter’ and the instrument big ‘pen’.
(68)
a.
b.
Damay
bind-e
bataaxal bi
big.
1SG.AFF write-APPL letter
DEF1 pen
‘I am writing the letter with a pen.’
!156
First, the verb merges with the theme object bataaxal bi ‘the letter’, defining the event
and forming the VP. Then ApplO merges with VP but does not merge up. Instead it
merges downward. Since the theme object is already in the complement of V position, the
merging of ApplO causes a reanalysis: both ApplO and the theme cannot occupy the
complement of V position simultaneously. The theme object ends up “inside” the ApplO
phrase. Then the applied object merges with ApplO. Again, the applied object merges
downward, which places the theme in the higher position in the phrase, the specifier. So
although the structure of ApplO looks very similar to Pylkkänen’s Low Appl, its
derivation is very different.
Locative applicatives involve the same head as instrumental applicatives, ApplO.
This explains why they exhibit the same morphology and object and c-command
properties. The derivation proceeds in the same fashion as the instrumental example in
(68).
(69)
a.
Ma ngi-y
jàng-e
téere bi
sama
1SG 1SG-IMP read-APPL book DEF1 1SG.POSS
‘I am reading the book at my house.’
kër.
house
b.
4.4.1. Lack of object scrambling
Now that the formation of instrumental and locative applicatives has been outlined, I
show how object and c-command properties are accounted for under this analysis. It was
!157
shown that the word order of the objects in instrumental and locative applicatives is
fixed. The theme always precedes the applied object, unlike benefactive and dative
applicatives, which allow object scrambling leading to variable word order of the objects.
Adopting the analysis presented here, the lack of object scrambling in instrumental and
locative applicatives results from the structure. Theme objects always precede oblique
applied objects in Wolof because there is no position available above the theme to which
the applied object could scramble. I explain this restriction in detail below.
In benefactive and dative applicatives, it was argued that the word order varies
because the lower object, the theme, scrambles to a position to the left of the applied
object, the second specifier of ApplT in benefactives or ApplE in datives. ApplT and
ApplE, as phase heads, were argued to have either one or two EPP features. These EPP
features target the theme object and cause it to raise to the Appl projection from its
original position in the VP. In instrumental and locative applicatives, the lower object is
already contained within the ApplO projection. Movement from its complement to its
specifier violates anti-locality (Grohmann 2003) and is thus not permitted. This results in
the fixed word order exhibited by instrumental and locative applicatives. Since theme
objects in benefactive and dative applicatives move to a new projection, ApplE or ApplT,
when they scramble, they do not violate anti-locality.
As with the structures we saw in chapter 3, the next step in the derivation is the
merger of v, which selects an external argument, the agent. In Wolof, v does not have
extra EPP features so there is no trigger for A-movement of the instrument or locative to
v. As for wh-movement, I assume EPP features which trigger A-movement to be different
than wh-features. In wh-movement, v inherits a wh-feature from C and this inherited
feature triggers movement of the wh-phrase to v and from there is attracted to C.
!158
4.4.2. Extraction
Turning to extraction, I address why the theme object, which is the high object, is
unavailable to wh-extraction and focus while the instrumental or locative object, which is
low, is available.
Instrumental - Instrument focused
(70) a.
Paakaa bi
laa
dagg-e
yàpp
knife
DEF1 1SG.OFOC cut-APPL meat
It was the knife with which I cut the meat.’
yi.
DEF.PL
Locative - Theme questioned
b.
*Lan nga togg-e
kër
Boris?
what 2SG cook-APPL house Boris
‘What did you cook at Boris’ house?’
‘With what did you cook at Boris’ house?’
Instrumental - Theme focused
(71) a.
*Yàpp yi
laa
dagg-e
paakaa
meat
DEF.PL 1SG.OFOC
cut-APPL knife
‘It was the meat that I cut with the knife.’
bi.
DEF1
Locative - Locative questioned
b.
Fan
nga (>Foo) togg-e
ceebu-jën?
where 2SG
cook-APPL ceebu-jën
‘Where did you cook ceebu-jën?’
In instrumental and locative applicatives, the theme cannot be extracted as evidenced by
the ungrammaticality of the sentences in (70) while the applied instrumental or locative
object can as seen in (71) (examples repeated from 4.2.2. for convenience).
The lower object can be A-bar extracted past the higher object as long as the
higher object does not have wh-features (McGinnis 2001). The higher object only acts as
intervener if it has a wh-feature. This explains why the lower object can be extracted past
the higher object as in (71). However, this does not explain, why a higher object with whfeatures cannot be extracted as seen in (70), illustrated without movement in (72).
!159
(72)
What blocks the NP lan from raising to CP through vP in (72) is not known.
Remember that the lower object, the locative, can move as seen in the sentences in (71b).
I tentatively suggest that the theme object is blocked in Wolof due to the nature of
downward merge and the reanalysis the theme undergoes. More evidence is needed to
make a firm claim about the extraction facts in instrumental and locative applicatives.
English shows a parallel pattern where the higher object, the dative, is blocked from Abar extraction.
In English, the dative object is the high object relative to the theme but cannot be
extracted for questioning or focus.4
(73)
a.
I gave Alex a book.
b.
*Who did you give a book?
c.
*It was Alex who I gave a book.
There are some English speakers who accept sentences like (73b) and (c) but there is a
general consensus5 that such sentences are worse than sentences with extracted theme
objects like in (74a) and (b). Remember that in English the theme is the lower object.
4
It cannot be extracted for syntactically marked focus like a cleft. It can be focused with a phonologically
marked focus placing an accent on the object while in situ. Since the discussion here is on extraction, the
accented focus is not relevant.
5
See Hudson 1992 for a more in-depth discussion about the debate surrounding the acceptability of
extraction of the dative object in English applicatives.
!160
(74)
a.
What did you give Alex?
b.
It was a book that I gave Alex.
The pattern where the high object is unavailable for extraction while the low
object is what is seen in Wolof locative and instrumental applicatives. There is currently
no good account in the literature for this phenomenon. For English, Hallman (2015) is
forced to stipulate the absence of movement from vP1, the higher vP in his structure while
movement can take place from vP2, the lower vP. He says this appears to be an specific
instance of the more general observation that A-bar movement targets lower elements
over higher elements.6 However, given Minimality, the opposite is expected; as an
intervener, the higher object should be targeted first provided it contains the correct
feature. In the case of Wolof and English, the higher object is clearly an intervener
making the fact that it can’t move while the lower object can even more surprising.
Additionally, assuming movement is blocked from the higher vP1 (analogous to Appl)
does not explain the extraction facts seen between benefactive and dative applicatives.
Either object can be extracted so we know movement is not blocked by the Appl phrase.
Thus, although the extraction restriction remains a mystery, it is a mystery that is present
elsewhere in applicatives cross-linguistically.
4.4.3. The in-between nature of instruments and locatives
The ApplO analysis proposed above reconciles the contradictory nature of instruments
and locatives. Recall from chapter 2 that cross-linguistically instrumental and locative
objects do not act entirely like VP-external objects (benefactives) nor do they behave
entirely like VP-internal objects (themes) (see chapter 2, section 2.2.2.). Changing the
instrument or locative object does not change the event described or the semantic role of
the theme object, whatever it may be. The ‘NP’ in both (75) and (76) is propelled through
the air regardless of how that action was initiated or where it occurred.
6
See Chomsky 1981, Huang 1982, and Lasnik and Saito 1984 for details on the preference of A-bar
movement of objects over subjects.
!161
(75)
(76)
a.
I threw NP with a ball launcher.7
b.
I threw NP with my hand.
a.
I threw NP in the park.
b.
I threw NP at the stadium.
Contrasting this with themes, we see that changing the theme object can change the event
and the relationship between the event and the other participants.
(77)
a.
I threw a ball at the arena.
b.
I threw a fight at the arena.
So instruments and locatives behave differently than themes in this regard. Now
comparing instruments and locatives to VP-external objects, like benefactives, shows
they don’t pattern with VP-external objects either. Instruments and locatives can
incorporate into adjectival passives, unlike benefactive objects.
(78)
a.
hand-made cookies
instrument
b.
spoon-fed children
instrument
c.
home-made cookies
locative
d.
*children-baked cookies
benefactive
e.
*boss-given flowers
goal/benefactive
(Marantz 1993:147)
Under the ApplO analysis, the instrument or locative is not selected by the verb and
merges with the VP, not V, so it doesn’t share the same behaviour as theme objects, which
are selected by the verb and merge with V. For example, because instruments and
locatives are not sisters of V, they are not involved in defining the event as themes are.
However, instruments and locatives do not behave like other VP-external objects, like
benefactives, because they end up within the minimal VP due to downward merge. Since
7 A ball
launcher is used to increase the throwing range of a ball, typically when playing with dogs.
!162
they are inside the domain of the verb, they can be incorporated into adjectival passives
and are intuitively part of the event.
4.4.4. Other applicative proposals
In this section I show that thematic and raising applicative structures are not able to
account for instrumental and locative applicatives. I also show Marantz’ multiple merge
orders for instruments and locatives fail to capture these constructions. The fact that these
analyses don’t explain these applicatives supports my proposal in which another
applicative head, which I argued to be ApplO, must be involved.8
Although useful in analyzing dative and benefactive applicatives as seen in
chapter 3, neither Raising nor Thematic Appl are able to account for instrumental and
locative in situ applicatives in Wolof. Looking at each structure, Thematic Appl would
seem the most likely choice for instrumental and locative applicatives because they are
compatible with intransitive predicates. Unfortunately, ApplT’s structure predicts the
wrong word order and c-command. As seen in (79), the instrumental (or locative) object
c-commands and precedes the theme object.
(79)
As we have seen, the theme precedes and c-commands the instrument. A second
specifier position on ApplT, which I used to explain variable word order in benefactive
8
Predicates which involve both instrumental and benefactive applicatives and the interaction of the
different applicative affixes may also shed light on the structure of applicative VPs. Such sentences do exist
in Wolof (see Buell and Sy 2005, 2006) but I leave this more complex issue to future research.
!163
and dative applicatives, cannot save the structure here. I argued in chapter 3 that ApplT
can have an extra EPP feature, which when present triggers movement of the theme to the
ApplT projection. Since the theme moves after the applied object has merged, it ends up
to the left of the applied object and c-commands it. One could argue that the theme applied word order in instrumental and locative applicatives results from the same
movement process of the theme to a second specifier. However, there are two problems
with such an analysis. Weak cross-over data (see section 4.3.2.) indicate the theme is
never in a position c-commanded by the instrument. We know this because the theme can
raise to the left for questioning without incurring a weak cross-over violation, meaning
the applied object does not c-command the trace of the theme. If the theme were
generated as the complement of V and always raised to the specifier of ApplT, its trace
would still be c-commanded by the applied oblique object and we would expect weak
cross-over effects to be present when the theme is questioned.
(80)
Thus, weak cross-over provide strong evidence against an analysis involving
ApplT and obligatory scrambling of the theme object. An object scrambling analysis for
instrumental and locative applicatives has a second problem. Since the theme never
appears to the right of the instrument or locative, the object scrambling would have to be
obligatory in the case of instrumental and locatives, but optional in the case of
benefactive and dative applicatives. There is no way to define ApplT as having two
!164
obligatory EPP features when combined with an instrument or locative while
simultaneously defining it as optionally having two EPP features when combined with a
benefactive.
Finally, a raising applicative analysis is untenable because instrumental and
locative applied objects are compatible with intransitive predicates while raising
applicatives are only compatible with predicates that encode a transfer of possession as
per Georgala (2012). A raising applicative structure is not possible for instrumental and
locative applicatives because of its expletive nature. It has no semantic content and
cannot assign a role to the instrument or locative. Given these objects can be omitted,
they can’t be selected by the verb like dative objects can.
Now that thematic and raising applicative analyses have been ruled out for
instrumental and locative applicatives in Wolof, I turn to Marantz’ (1993) proposal
according to which the instrument or locative can merge in two different positions: either
as the sister of V with the theme as the sister of Appl’ as in (81a) or as the sister of Appl’
with the theme as the sister to V as in (b).
(81)
a.
b.
The different merge orders result in the variable object orders seen in some languages,
like Chicheŵa and Chaga (see Marantz 1993). In these languages, unlike Wolof, either
object can be adjacent to the verb. When the theme precedes the instrument or locative,
the instrument merges first as in (81a). When the instrument or locative precedes and c-
!165
commands the theme, the theme merges first as in (81b). For Wolof, however, one would
have to stipulate that the theme never merges first.
The major drawback of Marantz’ analysis is that when the instrument merges first, as
illustrated in (81a), the theme is severed from the verb and introduced by Appl.
According to his proposed structure, the theme is now selected by the Appl head and lies
outside the minimal VP. Although the structure does capture the word order and ccommand facts of Wolof instrumental applicatives, severing the theme object is counterintuitive given the role the theme plays in defining the event. If the theme merges outside
the minimal VP, we lose the intuition about V+Theme complex that motivated Semantic
Compositionality in the first place (Marantz 1984, 1993).
The ApplO analysis captures the VP-internal aspects of instrumental and locative objects
- they end up inside the domain of the verb - while generating them at the VP level. The
theme stays selected by the verb in this analysis and appears within the minimal VP in
keeping with the notion a compositionally defined event. Finally, the theme c-commands
and precedes the instrument.
4.4.5. Movement out of VP
Before ending the discussion on instrumental and locative applicatives, I would like to
address the preference of fronted instrumental and locative objects over in situ
applicatives. Although fronted applicatives have not been the focus of this chapter, they
are considered more natural and are used far more commonly than in situ applicatives. As
mentioned previously, in situ applicatives are deemed marginal by Saint Louis speakers.
The main strategy for facilitating an instrumental or locative applied object is to focus it
so it appears in the left periphery outside the VP, described in section 4.1.4.9
9
See Torrence 2012 for a detailed analysis of focus and the left periphery in Wolof.
!166
Wolof instrumental applicative
(82) Paaka
bi
laa
dagg-e
yàpp.
knife
DEF1 1SG.OFOC
cut-APPL meat
‘It was with a knife that I cut the meat.’
Wolof locative applicative
(83) Kër
Boris
laa
togg-e
ceebu-jën.
house Boris
1SG.OFOC cook-APPL rice-fish
‘It was at Boris’ house where I cooked ceebu-jën.’
Why instrumental and locative applied objects resist staying in the VP is unclear. The
phenomenon, however, is not limited to Wolof.
Kiswahili instrumental applicatives resist two post-verbal objects like Wolof.
Regardless of word order, a Kiswahili applicative with both a post-verbal theme and
applied instrument is not grammatical, as shown in (85a) and (b). One strategy to allow
applied instrumental objects is to move the instrument to a topic position at the beginning
of the sentence, as seen with kisu ‘knife’ in (85c) (examples from Ngonyani 1998:81-83).
Kiswahili - Instrument > Theme
(84) a.
??wa-toto wa-li-vunj-i-a
ma-we
2-child
2-PST-break-APP-FV 6-rock
‘The children broke the pot with rocks.’
ch-ungu.
7-pot
Kiswahili - Theme > Instrument
b.
??wa-toto wa-li-vunj-i-a
ch-ungu ma-we.
2-child
2-PST-break-APP-FV 7-pot
6-rock
‘The children broke the pot with rocks.’
Kiswahili - Topicalized instrument
c.
ki-su,
wa-li-kat-i-a
nyama.
7-knife
2-PST-cut-APPL-FV
9.meat
‘The knife, they cut the meat with (it).’
!167
This is similar to the focus position applied instruments occupy in Wolof.
A another Bantu example of an applied instrumental object appearing outside the
VP is from Chicheŵa. The instrument must surface as the subject of a passive sentence
with the causative morpheme attached to the verb. If the instrument, khásu ‘hoe’, remains
an object in the VP instead of raising to subject position via passivization, then the
sentence is ungrammatical.
Chicheŵa instrumental
(85) a.
Khásu lí-ma-(li-)lim-its-ídw-a
hoe
SP-HABIT-(OP-)farm-CAUS-PASS-IND
‘The hoe is farmed with (by John).’
b.
*Jóni á-ma-(yi-)lemb-éts-a
John
SP-HABIT-(OP-)write-CAUS-IND
‘John writes with a pen.’
(ndí Jóni).
by John
péni.
pen
(Marantz 1984:245-246)
Looking outside of the Niger-Congo language family, several Austronesian
languages, also require the applied object to appear outside the VP, in the topic/subject
position in the case of Tagalog (McGinnis 2005) and Malagasy (Paul 2000).
In Tagalog the applied object moves out of the VP and into the subject position,
marked by the article ang and oblique topic morphology on the verb. If the instrument is
not marked as the subject, then the sentence is ungrammatical.
!168
Tagalog
(86) a.
b.
I-pinang-lakad ng
lalaki
ang
tungkod.
OBL-ASP-walk
CS1 man
ANG1 stick
‘The man walked with a stick.’
*Nag-lakad
ng
tungkod ang
lalaki.
NOM.ASP-walk CS1 stick
ANG1
man
‘The man walked with a stick.’
(Rackowski 2002 cited from McGinnis 2004:192)
Like Tagalog, Malagasy requires applied objects to appear in topic position which
places them outside the VP. In sentence (87a), there is no applied object. The benefactive
is contained in a prepositional phrase hoan’ny vehivavy ‘for the woman’. If the
benefactive object in an applicative structure is left in situ, as in (87b), the sentence is
ungrammatical. Circumstantial topic (CT), which makes the benefactive object the topic,
must be used for the applied object, ny vehivavy ‘the woman’ to be licit as shown in
(87c). This means that the benefactive ny vehivavy ‘the woman’ is the topic ,which is
analogous to subject position in Wolof. In the examples, the applied object is italicized
and the topic is underlined.
(87)
a.
Nahandro
ny
trondro hoan’ny
vehivavy ny
lehilahy.
PST.AT.cook DET1 fish
for.GEN.DET woman
DET1 man
‘The man cooked fish for the woman.’
b.
*Nahandro
PST.AT.cook
c.
ny
vehivavy
DEF1 woman
1
ny
trondro
DET1 fish
Nandrahoan’ny
lehilahy ny
trondo
PST.CT.cook.GEN.DET
man
DET1 fish
‘The man cooked fish for the woman.’
ny
lehilahy.
DET1 man
ny
vehivavy.
DET1 woman
(Paul 2000:120)
!169
The reason that applied objects are not allowed to surface within the VP in certain
languages is not well understood. The point of this section is simply to show that similar
behaviour is seen in applicatives cross-linguistically.
4.5. Conclusion
In this chapter, we have seen that instrumental and locative applicatives show different ccommand and object properties than benefactive and dative applicatives. Instrumental
and locative applicatives show a mix of symmetrical and asymmetrical properties. Only
the theme is allowed to appear adjacent to the verb which points to the theme as the true
direct object. However, the instrument or locative, and not the theme, can be extracted,
pointing to the applied object as having direct object status. As for pronominalization,
either object can be replaced with a clitic pronoun, but only one at a time. Sentences
where both objects are clitic pronouns are not grammatical. To contrast, benefactive and
dative applicatives, allow either object to appear adjacent to the verb, allow either object
to be extracted, and both objects can be pronominalized with clitics simultaneously.
C-command in locative and instrumental applicatives also differs from benefactive and
dative applicatives. In benefactive and dative applicatives, the object adjacent to the verb
c-commands the second object. In instrumental and locative applicatives, we’ve seen
using quantifier binding and weak cross-over that the theme asymmetrically c-commands
the applied instrumental or locative object. The applied object does not c-command the
theme.
I argued that a third Appl head, ApplO, is involved in the formation of instrumental and
locative applicatives. This head merges with the VP in a downward fashion following
McGinnis (2005). The downward merge places the applied instrument or locative to the
right of the theme and within its c-command domain explaining the word order and ccommand facts in Wolof.
!170
CHAPTER 5
5. CONCLUSION
In this thesis, I have investigated a particular piece of Wolof syntax: the structure of
applicative predicates and their properties. To this end, I presented a review of basic
properties of Wolof clauses and their syntax.
In chapter 2, I presented a typology of applicatives from different languages while
focusing on their object and c-command properties. Applicatives, rather than languages,
are classified as symmetrical or asymmetrical based on evidence from languages where
certain applicatives are symmetrical but other applicatives are asymmetrical. Several
major approaches to applicatives within the generative framework were presented and
although promising, they were unable to capture all types of applicatives crosslinguistically. I argued that a subtler approach hitherto was needed to account for the full
range of data in Wolof applicatives. I thus proposed a new Appl head, Oblique Appl for
instrumental and locative applicatives, in addition to adopting Georgala’s Thematic and
Raising Appls (2012) for benefactive and dative applicatives respectively.
In chapter 3, I showed that both benefactive and dative applicatives are symmetrical
applicatives and they share object properties and c-command configurations. Despite
these similarities, I also showed dative applicatives do not allow unspecified object
deletion which distinguishes them, along with morphology and semantic role, from
benefactive applicatives, which do allow unspecified object deletion. Due to these
differences, I conclude that benefactive and dative applicatives do not share the same
structure; they result from different derivations involving different Appl heads.
I argued that benefactive applicatives have a Thematic Applicative structure and datives
have a Raising Applicative structure. Benefactive objects are selected by the ApplT head
while dative objects are selected by the lexical verb and raise to the ApplE projection.
This approach to applicatives accounts for the syntactic similarities and object properties
!171
of benefactive and dative applicatives because the applied object is licensed in the same
position in both benefactives and datives. However, it also accounts for the differences in
selectional restrictions since the applied objects are selected in different ways. Different
morphology between the two types also falls out naturally since they involve different
heads and ApplT and ApplE have different morphological realizations.
As we saw in chapter 4, instrumental and locative applicatives cannot be accounted for
using Thematic or Raising Applicatives. They require a third type of applicative structure
which I call Oblique Applicative. Instrumental and locative applicatives show different ccommand and object properties than benefactive and dative applicatives. Instrumental
and locative applicatives show a mix of symmetrical and asymmetrical properties. Only
the theme is allowed to appear adjacent to the verb, which points to the theme as the true
direct object. However, the instrument or locative, and not the theme, can be extracted,
pointing to the applied object as having direct object status. As for pronominalization,
either object can be replaced with a clitic pronoun, pointing toward symmetrical
applicatives but only one object can be pronominalized at a time. To contrast, benefactive
and dative applicatives, which I argued to be symmetrical, allow either object to appear
adjacent to the verb, allow extraction of either object, and both objects can be
pronominalized with clitics simultaneously.
C-command in locative and instrumental applicatives also differs from benefactive and
dative applicatives. In benefactive and dative applicatives, the object adjacent to the verb
c-commands the second object: scrambling of the theme object to a position preceding
the applied object results in a change in c-command. In instrumental and locative
applicatives, I showed that the theme asymmetrically c-commands the applied
instrumental or locative object and the applied object cannot c-command the theme.
The object properties and c-command of instrumental and locative applicatives motivated
a more fine-grained approach to applicatives. I argued that a third Appl head, ApplO, is
involved in the formation of instrumental and locative applicatives. This head merges
!172
with the VP in a downward fashion following McGinnis (2005). The downward merge
places the applied instrument or locative to the right of the theme and within its ccommand domain, explaining the word order and c-command facts in Wolof.
In the ApplO analysis, I adopted the non-standard operation of Downward Merge
(McGinnis 2005). Given that I posit distinct positions for the distinct applicatives
(benefactive and dative merging upwards and instrumental and locative merging
downwards), further evidence and support for existence Downward Merge might come
by exploring the interactions between these affixes in verbs with multiple applicatives
(like the -al of benefactives and the -e of obliques). Such sentences are possible in Wolof
although I left these more complex issues for future research.
The analysis proposed here reconciles the presence of one merge site for Appl heads
while accounting for the different properties exhibited amongst the different type of
applicatives. It is argued that the three Appl heads, ApplT, ApplE, and ApplO all merge
above the lexical VP. The differences between the three types of applicatives arise from
selectional differences (raising applicatives and thematic applicatives) and the direction
of Merge (oblique applicatives).
Although I described multiple constructions that are used in Wolof to express
instrumental and locative object, I only proposed an analysis for in situ applicatives. I did
this because the in situ applicatives showed the most about VP structure in applicatives
which is the focus of this dissertation. I left the issue of hybrid applicatives, called noncanonical applicatives by Dione (2013) and Creissels (2004), for later study. These data
deserve further study as they have much to tell us about the augmentation of predicates.
Another issue that also merits further study is the extraction of objects in applicatives,
particularly in instrumental and locative applicatives. These data are not limited to Wolof,
but found in applicatives across diverse languages. The issue has been discussed in the
literature for many years and such data still resist formal explanation. Expanding the
conversation on extraction asymmetries to include languages typologically different than
!173
English can provide new insight to not only applicatives but syntactic movement in
general.
!174
REFERENCES
Aoun, Joseph and Yen-hui Audrey Li. 1989. Scope and constituency. Linguistic Inquiry
20:141-172.
Alsina, Alex and Sam A. Mchombo. 1993. Object asymmetries and the Chicheŵa applicative
construction. In Theoretical aspects of Bantu grammar, ed. Sam A. Mchombo, 17-45.
Standford: CSLI Publications.
Baker, Mark. 1988. Theta theory and the syntax of applicative constructions in Chicheŵa.
Natural Language and Linguistic Theory 6:353-389.
Baker, Mark. 1992. Thematic conditions on syntactic structures: Evidence from locative
applicatives. In Thematic Structure: Its Role in Grammar, ed. Iggy Roca, 23-46.
New York: Foris Publications.
Baker, Mark. 1996. On the structural position of themes and goals. In Phrase structure
and the lexicon, eds. Johan Rooryck and Laurie Zaring, 7-35. Dordrecht: Kluwer.
Barry, Boubacar. 2012. The kingdom of Waalo: Senegal before the conquest. New York:
Diaspora Africa Press.
Barss, Andrew and Howard Lasnik. 1986. A note of anaphora and double objects.
Linguistic Inquiry 17:347-354.
Beck, Sigrid and Kyle Johnson. 2004. Double objects again. Linguistic Inquiry
35:97-124.
Beck, Sigrid and William Snyder. 2001. The resultative parameter and restitutive ‘again’.
In Auditur vox sapientiae: A festschrift for Arnim von Stechow, eds. Caroline Féry
and Wolfgang Sternefeld, 48-69. Berlin: Akademie Verlag.
!175
Bobalijk, Jonathan. 2000. The ins and outs of contextual allomorphy. In The University of
Maryland Working Papers in Linguistics 10: 35-71. College Park, MD: University
of Maryland.
Boskovic, Zeljko. 1997. Superiority effects with multiple wh-fronting in Serbo-Croatian.
Lingua 102: 1-20.
Bresnan, Joan, and Lioba Moshi. 1990. Object asymmetries in comparative Bantu.
Linguistic Inquiry 21:147-185.
Bruening, Benjamin. 2001. QR obeys superiority: Frozen scope and ACD. Linguistic
Inquiry 32:233-273.
Buell, Leston and Mariame Sy. 2005. A fixed hierarchy for Wolof verbal affixes. BLS 31:
Special Session on Languages of West Africa, 25-36.
Buell, Leston and Mariame Sy. 2006. Affix ordering in Wolof applicatives and causatives.
In Selected proceedings of the 35th Annual Conference on African Linguistics:
African languages and linguistics in broad perspectives, eds. John Mugane, John
P. Hutchison, and Dee A. Worman, 214-224. Somerville, MA: Cascadilla
Proceedings Project.
Buell, Leston, Mariame Sy, and Harold Torrence. 2008. The syntax of Mirror Principle
violations in Wolof. Paper presented at International Morphology Meeting.
Vienna. Available at http://www.fizzylogic.com/users/bulbul/school/
BuellTorrenceSy-2008-handout.pdf.
Carlson, Greg. 1998. Thematic roles and the individuation of events. In Events and
grammar, ed. Susan Rothstein, 35-51. Boston: Kluwer Academic Publishers.
Chomsky, Noam. 1981. Lectures on government and binding. Dordrect: Foris.
!176
Chomsky, Noam. 1993. A minimalist program for linguistic theory. In The view from
building 20: Essays in linguistics in honor of Sylvain Bromberger, eds. Kenneth
Hale and Samuel Jay Keyser, 1-52. Cambridge, MA: MIT Press.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Chomsky, Noam. 1998. Minimalist inquiries: The framework. Cambrige, MA: MIT
Press.
Chomsky, Noam. 2000. The architecture of language, eds. Nirmalangshu Mukherji,
Subudhendra Narayan Patnaik, and Rama Kant Agnihotri. New Delhi and New
York: Oxford University Press.
Chomsky, Noam. 2013. Problems of projection. Lingua 130:33-49.
Creissels, Denis. 2004. Non-canonical applicatives and focalization in Tswana. In
Symposium on Syntax of the World’s Languages, Leipzig. 5-8 August. Available at
http://www1.ddl.ish-lyon.cnrs.fr/fulltext/Creissels/Creissels_2004d.pdf.
Creissels, Denis and Sylvie Nouguier-Voisin. 2004. The verbal suffixes of Wolof coding
valency changes and the notion of co-participation. Presented at Reciprocity and
Reflexivity, FU Berlin, 1-2 October. Available at: http://www.ddl.ish-lyon.cnrs.fr/
fulltext/Creissels/Creissels_2004e.pdf
Diesing, Molly. 1992. Indefinites: Linguistic Inquiry Monographs 20. Cambridge, MA:
MIT Press.
Diesing, Molly and Eloise Jelinek. 1993. The syntax and semantics of object shift.
Working Papers in Scandinavian Syntax 51:1-54.
Dione, Cheikh Bamba. 2013. Valency change and complex predicates in Wolof: An LFG
account. In Proceedings of the LFG13 Conference, eds. Miriam Butt and Tracy
Holloway King. CSLI Publications. Available at http://
!177
web.stanford.edu.proxy1.lib.uwo.ca/group/cslipublications/cslipublications/LFG/
18/papers/lfg13dione.pdf.
Diouf, Jean Léopold. 2002. Dictionnaire wolof-français. Dakar: Karthala Editions.
Dunigan, Melynda. 1994. On the clausal structure of Wolof. Doctoral dissertation, The
University of North Carolina at Chapel Hill.
Fal, Aram. 1999. Précis de grammaire fonctionnelle de la langue wolof. Dakar.
Georgala, Effi. 2012. Applicatives in their structural and thematic function: A minimalist
account of multitransitivity. Doctoral Dissertation, Cornell University.
Grewendorf, Günther. 2001. Multiple wh-fronting. Linguistic Inquiry 32:87-122.
Grimshaw, Jane. 1990. Argument structure. Cambridge, MA: MIT Press.
Grodzinsky, Yosef and Tanya Reinhart. 1993. The innateness of binding and coreference.
Linguistic Inquiry 24:69-101.
Grohmann, Kleanthes K. 2003a. German is a multiple wh-fronting language! In Multiple
wh-fronting, eds. Cedric Boeckx and Kleanthes K. Grohmann, 99-130.
Amsterdam: John Benjamins.
Grohmann, Kleanthes K. 2003b. Prolific domains: On the anti-locality of movement
dependencies. Philadelphia: John Benjamins Publishing Co.
Gropen, Jess, Steven Pinker, Michelle Hollander, and Richard Goldberg. 1991.
Affectedness and direct objects: The role of lexical semantics in the acquisition of
verb argument structure. Cognition 41:153-195.
Gruber, J. 1965. Studies in lexical relations. Doctoral dissertation, Massachusetts Institute
of Technology, Cambridge, Mass. Reprinted in Lexical Structures in Syntax and
Semantics, North-Holland, Amsterdam, 1976.
!178
Halle, Morris and Alec Marantz. 1993. Distributed Morphology and the pieces of
inflection. In The view from building 20, eds. Kenneth Hale and Samuel Jay
Keyser, 111-176. Cambridge: MIT Press.
Hallman, Peter. 2015. Syntactic neutralization in double object constructions. Linguistic
Inquiry 46:389-424.
Harley, Heidi. 2002. Possession and the double object construction. In Linguistic
Variation Yearbook 2, ed. Johan Rooryck, 31-70. Amsterdam: John Benjamins
Publishing Co.
Huang, 1982. Logical relations in Chinese and Theory of Grammar. Doctoral dissertation,
Massachusetts Institute of Technology.
Hudson, Richard. 1992. So-called ‘double objects’ and grammatical relations. Language
68:251-276.
Jackendoff, Ray. 1977. X-bar syntax: A study of phrase structure. Cambridge, MA: MIT
Press.
Jackendoff, Ray. 1987. The status of thematic relations in linguistic theory. Linguistic
Inquiry 18:369-411.
Ka, Omar. 1987. Wolof phonology and morphology: A non-linear approach. Doctoral
dissertation, University of Illinois at Urbana-Champaign.
Kayne, Richard. 1984. Correctedness and binary branching. Dordrecht: Foris.
Kimenyi, Alexandre. 1980. A Relational Grammar of Kinyarwanda. Berkley: University
of California Press.
Kimenyi, Alexandre, 1995. Kinyarwanda applicatives revisited. Paper presented at the
8th Niger-Congo Syntax Semantics Workshop. Boston University, 25–26 October,
!179
1995. Available at http://www.kimenyi.com/kinyarwanda-applicativesrevisited.php
Kirparsky, Paul. 1987. On theory and interpretation. In Linguistics and Literary Theory,
ed. F. Newmeyer. Manchester: Manchester University Press.
Kratzer, Angelika. 1996. Severing the external argument from its verb. In Phrase
structure and the lexicon, eds. Johan Rooryck and Laurie Zaring, 109-137.
Dordrecht: Kluwer Academic Publishers.
Larson, Richard K. 1988. On the double-object construction. Linguistic Inquiry
19:335-391.
Larson, Richard K. 1990. Double objects revisited: Reply to Jackendoff. Linguistic
Inquiry 21:589-632.
Lasnik, Howard and Mamoru Saito. 1984. On the nature of proper government.
Linguistic Inquiry 15:235-289.
Lewis, M. Paul, Gary F. Simons, and Charles D. Fennig, eds. 2014. Wolof. In
Ethnologue: Languages of the World, Seventeenth edition. Dallas, Texas: SIL
International. Online version: http://www.ethnologue.com. Accessed Dec. 5,
2014.
Marantz, Alec. 1984. On the nature of grammatical relations. Cambridge, MA: The MIT
Press.
Marantz, Alec. 1993. Implications of asymmetries in double object constructions. In
Theoretical aspects of Bantu grammar 1, ed. Sam A. Mchombo, 113-151.
Stanford: CSLI Publications.
Martinovic, Martina. 2013. Wh-morphology and cyclicity in Wolof. Available at http://
ling.auf.net/lingbuzz/001874/current.pdf?_s=QpMDJutiHFYSZjvi
!180
Matthewson, Lisa. 2004. On the methodology of semantic fieldwork. International
Journal of American Linguistics 70:369-415.
McLaughlin, Fiona. 2004. Is there an adjective class in Wolof? In Adjective classes: A
crosslinguistic typology, eds. R.M.W. Dixon and Alexandra Y. Aikhenvald
242-262. Oxford: Oxford University Press.
McCloskey, James. 1991. Clause structure, ellipsis and proper government in Irish.
Lingua 85:259–302.
McGinnis, Martha. 2001. Phases and the syntax of applicatives. In NELS 31, eds. Minjoo
Kim et Uri Strauss, 333-349. Amherst: University of Massachusetts.
McGinnis, Martha and Donna B. Gerdts. 2003. A phase-theoretic analysis of
Kinyarwanda multiple applicatives. In Proceedings of the 2003 Annual
Conference of the Canadian Linguistic Association, 154-165. Available at http://
homes.chass.utoronto.ca/~cla-acl/2003/McGinnis-Gerdts
McGinnis, Martha. 2005. UTAH at Merge: Evidence from multiple applicatives. MIT
Working Papers in Linguistics 49:183-200.
Nakanishi, Kimiko. 2001. Frozen scope and weak crossover in ditransitives. In Formal
Approaches to Japanese Linguistics 3: MIT Working Papers in Linguistics 41,
eds. M. C. Cuervo, D. Harbour, K. Hiraiwa, and S. Ishihara. Cambridge:
MITWPL.
Ngonyani, Deo. 1996. The morphosyntax of applicatives. Doctoral dissertation, UCLA.
Ngonyani, Deo. 1998. Properties of applied objects in Kiswahili and Kinendeule. Studies
in African Linguistics 27:67-95.
Ngonyani, Deo and Peter Githinji. 2006. The asymmetric nature of Bantu applicative
constructions. Lingua 116:31-63.
!181
Njie, Codu Mbassy. 1982. Description syntaxique du wolof de Gambie. Dakar: Nouvelles
éditions africaines.
Nouguier-Voisin, Sylvie. 2002. Relations entre fonctions syntaxiques et fonctions
sémantiques en wolof. Doctoral dissertation, Université Lumière Lyons 2.
Otani, Kazuyo. and John Whitman. 1991. V-raising and VP-ellipsis. Linguistic Inquiry
22:345-358.
Oyharçabal, Beñat. 2010. Basque ditransitives. In Argument structure and syntactic
relations: A cross-linguistic perspective, eds. Maia Duguine, Susana Huidobro,
and Nerea Madariaga, 233-260. Philadelphia: John Benjamins.
Payne, Thomas. 1997. Describing morphosyntax. Cambridge: Cambridge University
Press.
Paul, Ileana. 2000. Malagasy clause structure. Doctoral dissertation, McGill University.
Paul, Waltraud and John Whitman. 2010. Applicative structure and Mandarin
ditransitives. In Argument structure and syntactic relations: A cross-linguistic
perspective, eds. Maia Duguine, Susana Huidobro and Nerea Madariaga, 261-282.
Philadelphia: John Benjamins.
Phillips, Colin. 2003. Linear order and constituency. Linguistic Inquiry 34:37-90.
Polinsky, Maria. 2013. Applicative constructions. In The World Atlas of Language
Structures Online, eds. Matthew S. Dryer and Martin Haspelmath. Leipzig: Max
Planck Institute for Evolutionary Athropology.. Available at http://wals.info/
chapter/109, Accessed on 2015-08-08.
Pylkkänen, Liina. 2002. Introducing arguments. Doctoral dissertation, Massachusetts
Institute of Technology.
!182
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
Reinhart, Tanya. 1976. Quantifier scope: How labor is divided between QR and Choice
Functions. Linguistics and Philosophy 20:335-397.
Richards, Norvin. 1999. Feature cyclically and ordering of multiple specifiers. In
Working Minimalism, eds. Samuel D. Epstein and Norbert Hornstein, 127-158.
Cambridge, MA: MIT Press.
Richards, Norvin. 2002. Very local A-bar movement in a root-first derivation. In
Derivation and exploration in the Minimalist Program, eds. Samuel D. Epstein
and T. Daniel Seely, 227-248. Malden, MA: Blackwell Publishing.
Roberge, Yves and Sarah Cummins. 2005. A modular account of null objects in French.
Syntax 8:44-64.
Robert, Stéphanie. 1991. Approche énonciative du système verbal: Le cas du Wolof.
Paris: Editions du CNRS.
Russell, Margaret. 2007. The syntax and placement of Wolof clitics. Doctoral
dissertation, University of Illinois at Urbana-Champaign.
Schwartz, Arthur. 1975. Wolof: A language with no direct object. The Journal of West
African Linguistics 10:219-267.
Sportiche, Dominique. 1988. A theory of floating quantifiers and its corollaries for
constituent structure. Linguistic Inquiry 19:425-449.
Synder, William. 2001. On the nature of syntactic variation: Evidence from complex
predicates and complex word-formation. Language 77:324-342.
Tamba, Khady, Harold Torrence, and Malte Zimmerman. 2012. Wolof Quantifiers. In
Handbook of quantifiers in natural language, eds. Edward L. Keenan and Denis
Paperno, 891-940. New York: Springer.
!183
Taylor, Jill. 1995. Code switching in Dakar. La Linguistique 31:63-78.
Thráinsson, Höskuldur. 2007. The syntax of Icelandic. Cambridge: Cambridge University
Press.
Torrence, Harold. 2012. The clause structure of Wolof: Insights into the left periphery.
Philadelphia: John Benjamins Publishing Company.
Zeller, Jochen and Jean Paul Ngoboka. 2006. Kinyarwanda locative applicatives and the
Minimal Link Condition. Southern African Linguistics and Applied Language
Studies, 24:101-124.
Zribi-Hertz, Anne and Lamine Diagne. 2002. Clitic placement after syntax: Evidence
from Wolof person and locative markers. Natural Language and Linguistic
Theory 20: 823-884.
!184
Curriculum Vitae
Name:
Christen Harris Post-secondary
Education and
Degrees:
Utah State University
Logan, UT, USA
2005 - 2009 B.A.
Western University
London, Ontario, Canada
2009 - 2015 Ph.D.
Honours and
Awards:
Western Graduate Research Scholarship
Western University
2009 - 2014
Graduate Research Award
Western University
2013 - 2014
Graduate Thesis Research Award
Western University
2013 - 2014
Related Work
Experience
Teaching Assistant
Western University
2009 - 2014
Publications and Presentations:
Harris, Christen. 2014. “Applicative Theory Applied to Wolof.” Presentation at The
Annual Conference on African Linguistics 45. University of Kansas. 17-19 April.
Harris, Christen. 2014. “Describing Wolof applicatives and theoretical implications.”
Invited speaker at Linguistics Talks at Western. Western University Canada. 6
March.
Harris, Christen. 2012. “Contexual Allomorphy in the Wolof Past Tense.” Presentation at
Canadian Linguistics Association Annual Conference. Wilfrid Laurier University.
26-28 May.
!185
Harris, Christen. 2011. “Le système focus-topique en Wolof.” Presentation at L’exemple
et le contre-example en littérature et en linguistique. Western University Canada.
21-22 October.
Harris, Christen. 2011. “The Structural Category of Past Tense (w)oon in Wolof.”
Working Papers of the Linguistics Circle of the University of Victoria 21, 34-42.
Harris, Christen. 2011. “The Structural Category of Past Tense (w)oon in Wolof.”
Presentation at 27th Northwest Linguistics Conference. University of Victoria.
19-20 February.
Andrighetti, Michelle, Roronhiakehte Deer, Christen Harris, Caelan Marr, Andrew
McKishnie, Philbert Musanganya, and Ileana Paul. 2010. “Is it a cleft? What
constructions can tell us about cross-linguistic differences.” Presentation at The
14th Bilingual Workshop in Theoretical Linguistics. University of Toronto. 8-9
December.
Paul, Ileana, Michelle Andrighetti, Roronhiakehte Deer, Christen Harris, Caelan Marr,
Andrew McKishnie, and Philbert Musanganya. 2010. “Is it a cleft? What
constructions can tell us about cross-linguistic differences.” Poster presentation at
Canada Research Chairs Thinking Ahead for a Strong Future. Toronto, Canada.
24-25 November.
Harris, Christen. 2009. “Searching for Identity in African Francophone Films.”
Presentation at Utah Academy of Sciences, Arts and Letters Annual Conference.
Brigham Young University. 10 April.
Harris, Christen. 2009. “Generational Differences Between North African Francophone
Literatures: The New Stories of Immigrants in France.” Undergraduate Honors
Theses. Paper 6. http://digitalcommons.usu.edu/honors/6.
Harris, Christen. 2009. “Generational Differences Between Beur and North African
Francophone Literatures: The New Stories of Immigrants in France.” Presentation
at Utah State University Student Showcase. Utah State University. 25 April.
Harris, Christen. 2008. “Searching for Identity in African Francophone Films.”
Presentation at Utah State University Languages, Philosophy and Speech
Communication Annual Symposium. Utah State University. 14 March.